Geiyecapixie

Wallet: 0xcae693bcf9696a2ebf0a62de767719b45f354f85 Window: 2026-05-03 to 2026-05-30 (28 calendar days, 28 active) Universe: 11,301 trades across 199 markets / 86 events · $2.49M gross turnover Net cash-flow P/L: +$68,127 on $1,211,928 deployed = +5.62% ROI in 28 days

The Punchline

Geiyecapixie is a live League of Legends bettor who watches LPL and LCK matches in real time, takes large directional positions on the in-game leader, and actively exits via the SELL engine rather than holding to resolution. Every trade in the sample is a LoL game-winner or series-winner market. No other game, no crypto, no politics - 100% LoL esports.

The strategy is a hybrid of two components. First, directional in-game betting: watching the live score, gold difference, and objective control, then buying the leading team's shares at market when a decisive move occurs. Second, same-event hedging: simultaneously holding small positions on the opposing side of the same match (or adjacent game within the same series) to recover capital if wrong, and selling that hedge aggressively into the orderbook when the primary position confirms.

The economics are steady rather than explosive. $1.21M of BUY notional across 28 days returns +$68,127 in realized P/L - a +5.62% monthly ROI. That sounds thin, but it represents a genuine, reproducible edge in a niche where most bettors either pick sides blindly or copy public favorites. The wallet's dominance at 3.0x+ conviction (87.5% dominant-side win rate across 72 markets) is the smoking gun: this trader knows who is winning the game before the odds catch up.

---

What He Trades

The universe is one game, multiple circuits:

League of Legends - LPL (China):  majority of volume
  Team slugs: BLG, WE, LNG, JDG, IG, WBG, NiP, TT, OMG, EDG, AL
League of Legends - LCK (Korea):  substantial minority
  Team slugs: T1, DK, HLE, FEARX, NongShim, DN SOOPers
LoL International: Esports World Cup qualifiers, cross-region
Market types: Game N Winner (single game), BO3/BO5 series winner

All 199 markets and all 86 events are LoL. The category classifier flags everything as "Other" because the standard keyword set does not include esports team names for LoL. The slug patterns confirm the universe: every market slug follows lol-[team1]-[team2]-[date]-gameN or lol-[team1]-[team2]-[date] (for series).

The top 10 markets by volume tell the competitive landscape:

The NiP vs BLG Game 1 market is the single most profitable trade at +$12,464 on $56K volume (+22.2% ROI on that market alone). BNK FEARX vs T1 Game 1 is the single worst at -$5,799. Both involve T1, which is historically a difficult market to read because the crowd-driven T1 popularity creates persistent overpricing on T1 outcomes.

---

The Order of Operations - One Match, Trade by Trade

This trace covers the BLG vs Team WE - Game 4 Winner market on 2026-05-30 (lol-blg-we-2026-05-30-game4), which resolved Team WE as the winner. The CSV provides complete coverage. This game was part of a series where BLG had led 2-1, making Game 4 a must-win for WE.

Walk-through of what this tells us:

The game starts close. At 13:16:56, WE is a 30% underdog and the wallet opens a $612 BUY on WE. Within 47 seconds it sells most of that position back (13:17:03 and 13:17:46) at $0.33-$0.37, near breakeven with a slight loss - the price didn't move as expected and he cuts.

Then at 13:18:20 something happens in the game (most likely a WE teamfight win or baron takedown). He re-enters WE at $0.44 with 2,000 shares ($890), then sells aggressively into the price rise at $0.40-$0.43 - essentially a scalp. At 13:20:22 he goes much bigger on WE: 3,700 total shares at $0.41-0.48 ($1,667 total). Within 2 minutes (13:22:19-23) the price has risen to $0.61 and he sells the entire position for $2,235, netting +$568 on the flip.

The complexity escalates. He then buys WE again at $0.84 for $1,692 (13:28:11), suggesting another decisive WE move. He then simultaneously starts buying BLG at $0.66-$0.93 - the opposing side - between 13:34 and 13:39, creating a both-sides position. He sells the BLG back at $0.97 for +$1,711, converting that hedge into profit when BLG momentarily threatened a comeback. Finally, the terminal move: at 13:40:40 through 13:40:51 he pumps $8,690 into WE shares at $0.78-$0.99, correctly anticipating the close-out. The 10,000-share SELL at $0.999 at 14:01:42 is the final cash-out before resolution.

This is an expert live LoL bettor. The trade sequence follows the exact rhythm of a close BO3/BO5 series game: early uncertainty, decisive momentum shift, partial fade, second decisive shift, and final blowout. Each move corresponds to an in-game event.

---

Why It Works - The Math

The strategy's positive EV comes from information advantage, not price-structure advantage. The paired cost of $1.27 across 115 both-sides markets means he is paying $0.27 over $1.00 for both sides combined - no guaranteed spread profit here. The profit is entirely directional.

<pre><code>Dominant-side win rates by conviction tier: 1.0-1.5x dominance: 72.2% (18 markets) 1.5-2.0x dominance: 90.9% (11 markets) ← highest accuracy 2.0-3.0x dominance: 57.1% (14 markets) ← worst tier, possibly catch-up sizing 3.0x+ dominance: 87.5% (72 markets) ← workhorse tier

Overall resolved win rate: 67.7% on 7,037 resolved BUYs Average entry price (capital-weighted): ~$0.80 (majority in $0.80-$1.00 bands) Most capital in $0.90-$1.00 band: $360,764 (29.8% of capital, 92.2% WR)

EV calculation per dollar at the $0.90-$1.00 band: WR = 92.2% → expected shares paid at $1.00 = 0.922 / $avg_price(~$0.95) = 0.970 shares per $1 Expected payout = 0.970 shares × $1.00 = $0.970 + ...actually resolves at 92.2% * (1/0.95) = $0.970 ROI at this band: +5.46% (from data)

EV calculation at the $0.40-$0.50 band: WR = 52.2% → expected payout = 0.522 / $0.45 avg = $1.16 per $1 deployed ROI at this band: +10.1% (from data) - BEST ROI band</code></pre>

The best ROI per dollar is actually at $0.40-$0.50 entries (+10.1%), meaning the mid-game entries where he identifies value at coin-flip prices return more per dollar than the near-certain high-price entries. But the volume-weighted capital allocation is toward $0.70-$1.00 because those are the confirmatory enters where he has strong conviction and wants large exposure.

---

Phase 1 - Trader Profile

Scale and Activity

Trade Size Distribution

The size distribution is deeply bimodal. The median of $6.16 reflects hundreds of small probe trades ($3-10 each) during ongoing games. The mean of $220 reflects the handful of large conviction clips ($1,000-$5,000) that carry the capital. The max of $17,082 is a single close-out SELL. The top 5% of trades carry 53.5% of total notional. This is power-law capital concentration with small probes and large decisive fills.

The Lorenz curve in the data confirms extreme inequality:

• Bottom 50% of trades: 0.9% of capital
• Bottom 85% of trades: 9.5% of capital
• Top 5% of trades: 53.5% of capital
• Top 1% of trades: 86.1% of capital (via the Lorenz data: 1-0.860758 = 13.9% at P99, so top 1% carries 1-0.860758 of capital... actually the lorenz shows cumulative capital from bottom, so bottom 99% holds 86.1%, top 1% holds 13.9%)

Trading Hours (UTC)

Zero activity from 00:00-04:00 UTC. Peak window is 08:00-14:00 UTC (China Standard Time 16:00-22:00 = LPL primetime, Korea Standard Time 17:00-23:00 = LCK evening). The hardest 4 worst hours are 6, 7, 8, 14 UTC per the filter data, corresponding to early warm-up matches and late close-out games. The 12:00 UTC hour is the single biggest absolute P/L hour (+$23,358 on 1,565 trades), corresponding to ~20:00 CST - late LPL primetime with the most consequential matches.

Archetype

Semi-automated probe system with manually triggered large clips. Real-time game state information drives conviction. Both-sides participation (57.8%) is structural hedging rather than spread capture.

---

Phase 2 - Core Strategy Identification

Both-sides participation rate: 57.8% (115 of 199 markets had both YES and NO sides bought).

This is well above the 30% threshold for both-sides classification but the paired cost tells the real story: median paired cost of $1.2678, mean of $1.2733. Both sides consistently cost $1.27 for the pair, guaranteeing a $0.27 loss if held to resolution without exits. This is not spread capture. The trader is paying a premium for optionality - the ability to sell the losing side into the market before resolution, recovering $0.20-$0.40 on the dollar on the hedge leg.

The dominant-side classification by conviction bucket:

• 3.0x+ dominance: 72 markets, 87.5% dominant-side win rate - this is the core directional edge
• 1.5-2.0x: 11 markets, 90.9% win rate - highest accuracy tier (small sample but strong)
• 1.0-1.5x: 18 markets, 72.2% win rate - lower conviction, balanced books
• 2.0-3.0x: 14 markets, 57.1% win rate - anomalous underperformance (may be catch-up sizing on trailing games)

The strategy is definitively B (Directional Betting) with E (Hedged/Insurance Structure) overlay. The both-sides participation is insurance against being wrong, not an attempt to lock in guaranteed spread profit.

---

Phase 3 - Dominance Ratio Analysis

The dominance ratio analysis is the most informative phase for this trader:

The 3.0x+ tier carries the weight of the strategy. 72 markets with 87.5% dominant-side win rate is not a statistical accident at this sample size. This is persistent informational accuracy.

The 2.0-3.0x tier's 57.1% win rate is the one anomaly. These may be markets where the wallet is scaling into a conviction it held strongly but the game went the other way (the "catch-up sizing" problem - adding to a losing position mid-game). The BNK FEARX vs T1 Game 1 (-$5,799) is likely in this bucket: a large directional bet on the underdog that didn't materialize.

The second-side lag median of 544 seconds (~9 minutes) confirms this is intentional in-game hedging: the wallet opens the primary side first, watches the game for 9 minutes, then adds the opposing side as a hedge when the game state warrants it.

---

Phase 4 - Entry Price Analysis

The capital distribution peaks in the $0.90-$1.00 band ($360K, 29.8% of capital) and $0.80-$0.90 band ($289K, 23.9%), together comprising 53.7% of capital. This is the heavy-favorite zone where the wallet buys confirmatory positions once a team has a decisive lead. The win rates (92.2% and 76.8%) match the prices paid - the market is well-calibrated and the wallet is not sniping mispricings here. The ROI (~5.5% across both bands) is consistent with the overall book.

The ROI peak at $0.40-$0.50 (+10.1%) represents the alpha-dense midgame entries - moments when one team has just taken control but the orderbook hasn't fully repriced yet. These entries produce the highest return per dollar because the price still carries uncertainty premium that the wallet's live read has resolved.

The sub-bucket inspection shows no single-cent anchor. The wallet enters at every price from $0.20 to $0.99 depending on the game state, which is consistent with an operator reacting to live events rather than targeting a specific price level.

---

Phase 5 - Category Breakdown

The entire book is "Other" in the standard classifier. The real breakdown is by competition:

The best individual markets by P/L are concentrated in LPL Group Stage and LCK Playoffs, where match schedules are dense (multiple games per day) and the wallet can trade 10+ markets per afternoon session.

Assessment: the single "Other" category is Elite by ROI standard (>5% with large resolved count) and the entire book is within this vertical.

---

Phase 6 - Timing and Execution

Burst Patterns

The CSV shows two distinct fill-size modes:

1. Probe mode: Repeated $3-5 fills (5 shares at $0.61, $0.63, $0.65, $0.67, etc.) spaced 10-60 seconds apart. These are the orderbook monitoring pings - small fills that let the wallet track price movement and test liquidity. The TT vs LNG Game 1 trace shows 30+ consecutive $3-5 fills across 15 minutes.

1. Conviction mode: Large single fills ($500-$5,000) clustered within 5-10 seconds of each other. These are the decisive position-builds triggered by in-game events. The BLG vs WE Game 4 trace shows $1,693 + $475 + $107 + $71 + $41 in 32 seconds (13:28:11 to 13:28:23) - a clear rapid fan-out through the orderbook.

The median inter-fill gap of 5 seconds and 70.3% of fills under 10 seconds confirm bot-assisted execution for at minimum the probe fills. The large clips are likely manual or semi-manual triggers.

Accumulation Window

The wallet does not return to markets after resolution. Each game is a single event with entry, management, and exit within the game duration (typically 30-50 minutes for an LPL game).

Day-of-Week Performance

Sunday is the dominant day (+$25,448, +10.8% ROI). Monday follows (+$9,361, +7.6%). Thursday is the only negative-ROI day (-0.5%). Weekend concentration is consistent with LPL and LCK scheduling: the most important, highest-stakes matches fall on weekends when the wallet's informational edge is presumably sharpest (more motivated engagement, marquee matchups).

Wednesday's unusually low win rate (52.8%) despite positive P/L suggests many mid-week matches were close games that resolved correctly but barely.

---

Phase 7 - Filter Experiments

Full filter analysis is in the Filter Strategy tab. Summary:

The high-conviction filter (dom 2x+) is the one meaningful lift: ROI improves from +5.62% to +6.72% (+1.1pp) while cutting capital deployed to $596K. This is the most actionable filter finding - skipping the 1.0-1.5x tier (where paired cost is highest and accuracy lowest) and focusing on the 2x+ conviction markets improves efficiency without dramatically reducing opportunity.

---

Phase 8 - Rolling Window Consistency

Every rolling 15-day window is green. The weakest 7-day window still prints +$6,648 (the May 10-16 period after a strong first week). The strategy is remarkably consistent: five weeks of production, no losing weeks, no catastrophic drawdown period.

The trajectory shows late-window acceleration: Week 21 (May 18-24) was the best at +$20,039, and the 15-day rolling windows peak at +$39,064 around May 23-24. This could reflect improved match quality (playoff rounds) or increasing position sizing as the wallet operator gained confidence.

The weekly win rate varied from 63.1% (Week 20) to 77.9% (Week 18, opening week). Week 20's lower win rate may correspond to more contested Group Stage matches with less decisive game states.

---

Phase 9 - P/L Decomposition

The decomposition reveals the cost of the hedging program: -$100,641 in structural spread loss (paying $1.27 for paired positions that structurally settle at $1.00 if held). This is offset by the active SELL engine recovering value from the losing-side positions before resolution. The $244,287 hedge tax figure represents the capital allocated to non-dominant sides in both-sides markets - almost all of which gets sold back into the market rather than settling at zero.

The net result is that the hedging program is approximately break-even on its own: the $100,641 structural loss is partially recovered by SELL exits on the hedge legs. The positive $68,127 P/L comes primarily from the dominant-side wins at 87.5%+ accuracy on high-conviction markets.

---

Phase 10 - Strategy Specification

One-sentence summary: A live LoL esports bettor who takes large directional positions on the in-game leader, holds a smaller opposing hedge for capital recovery, and actively exits both sides via the SELL engine rather than settling to resolution.

Edge source: Real-time game state information - the operator watches matches live and identifies leadership transitions (baron kills, teamfight wins, gold leads) before the Polymarket orderbook reprices. The 87.5% dominant-side win rate at 3.0x+ conviction is the quantified version of this informational edge.

What works: 3.0x+ dominance conviction tier (87.5% win rate). Sunday and Saturday (highest-stakes matches). 09:00-13:00 UTC (LPL primetime). $0.40-$0.50 entry band (best ROI). NiP vs BLG matchups (most profitable single market, +$12,464).

What drags: 2.0-3.0x conviction tier (only 57.1% win rate - the wallet's soft spot). Hour 6-7 UTC (early morning, lower-stakes matches). BNK FEARX vs T1 Game 1 (-$5,799 single worst market, possibly the most crowd-distorted market in Korean LoL).

Full implementation spec in the Playbook tab.

Wallet: 0xcae693bcf9696a2ebf0a62de767719b45f354f85 Window: 2026-05-03 → 2026-05-30 (28 active / 28 calendar days) Methodology: Cash-flow P/L = -buy_usdc + sell_usdc + remaining_share_payout. Resolved shares settle at $1 (win) / $0 (loss); open positions marked at last price.

Phase 1 - Trader Profile

Scale

Trade-size distribution (USDC per fill)

Inter-trade gap, same (market, outcome)

Phase 2 & 3 - Both-Sides Participation, Dominance Curve

• Both-sides rate: 57.79% (115 of 199 markets)
• Median paired cost: $1.2678
• Mean paired cost: $1.2733
• Paired cost % under $1.00: 8.7%
• Paired cost % under $0.97: 6.1%
• Median 2nd-side hedge lag: 544s

Dominance buckets

Phase 4 - Entry-Price Analysis

Phase 5 - Category & Vertical Breakdown

Phase 6 - Timing & Execution

Net P/L by hour (UTC)

Phase 8 - Rolling Window Consistency

• Rolling 7-day windows green: 28 of 28 (100.0%)
• Rolling 7-day P/L range: +$6,649 → +$25,055
• Rolling 15-day windows green: 28 of 28 (100.0%)
• Rolling 15-day P/L range: +$15,202 → +$40,744

Weekly P/L

Phase 9 - P/L Decomposition

Phase 10 - Top Markets by Volume

Top 10 winners by P/L

Top 10 losers by P/L

Report generated 2026-06-02 08:17 UTC.

Wallet: 0xcae693bcf9696a2ebf0a62de767719b45f354f85 Window: 2026-05-03 to 2026-05-30 Baseline: 7,037 resolved BUYs · 67.7% WR · $1,211,928 deployed · +$68,114 P/L · +5.62% ROI

---

The headline result

One filter produces genuine, meaningful lift. Most others are destructive or irrelevant.

The high-conviction dominance filter (dom 2x+, dominant side only) raises ROI from +5.62% to +6.72% while cutting capital deployed roughly in half. This is real improvement and reflects the genuine weaknesses in the low-dominance bucket (1.0-1.5x at 72.2% win rate) which the filter correctly excludes.

The price-band filter is modestly neutral at baseline but cuts absolute P/L by $49,500 - not because it destroys a profitable zone but simply because it excludes most of the capital. The hour filter produces a small genuine improvement. The category filter is identity-equivalent.

The most important finding is structural: the 2.0-3.0x dominance bucket underperforms (57.1% win rate) while the 3.0x+ bucket outperforms (87.5%). A modified filter that skips the 2.0-3.0x tier specifically and includes only 3.0x+ (plus the 1.5-2.0x tier which also outperforms at 90.9%) would produce even better ROI than the simple "dom 2x+" cut, though we cannot compute that precisely from the data provided.

---

Filter results table

---

Filter-by-filter commentary

1. Price band ($0.30-$0.70) → NEUTRAL-SLIGHTLY-POSITIVE ROI, DESTRUCTIVE ABSOLUTE

Applying the standard sweet-spot filter shifts ROI marginally up from +5.62% to +5.76% (+0.14pp) while cutting absolute P/L by $49,508. The ROI improvement is essentially noise at this scale. The filter cuts 4,528 trades out of 7,037 - including the entire $360,764 invested in the $0.90-$1.00 band, which returns +5.46% ROI (below filter improvement, so the filter does help slightly). It also cuts the $289,377 in the $0.80-$0.90 band (+5.78% ROI, nearly identical to filtered).

What the filter cannot do is separate the hedge legs from the conviction legs. When the wallet buys the underdog at $0.10-$0.20 (to hedge), those small-sized trades have terrible ROI and the filter correctly excludes them. But when it buys the favorite at $0.85-$0.99, those are the large profitable dominant legs and the filter incorrectly excludes those too.

The conclusion: the price filter is slightly ROI-positive by accident (it excludes the high-paired-cost hedge legs in the $0.10-$0.30 zone) but the improvement is too small to be meaningful and the absolute P/L destruction is severe.

For a replicator: do not apply a price filter to this strategy. The strategy legitimately operates across the full price spectrum because the game-state information changes the fair value continuously. A $0.85 buy on a team that just won a teamfight is better value than a $0.45 buy on a team that just lost one.

2. High-conviction filter (dom 2x+, dominant side only) → MEANINGFUL LIFT

This is the one filter that produces genuine, defensible improvement. Restricting to markets where the wallet allocated at least 2x more capital to one side and betting only the dominant side:

• Win rate rises from 67.7% to 82.4% (+14.7pp)
• ROI rises from +5.62% to +6.72% (+1.1pp)
• Capital deployed drops from $1.21M to $596K (-51%)
• Absolute P/L drops from $68,114 to $40,033 (-$28,081)

The ROI improvement is real because it filters out the 1.0-1.5x tier (72.2% win rate, $1.359 paired cost - the most "balanced" book with the highest insurance premium). By requiring at least 2x dominance, you ensure you're only backing the markets where the wallet has genuine strong conviction.

Caveat: the 2.0-3.0x bucket (57.1% win rate) contaminates the dom 2x+ filter. A refined cut at 3.0x+ would exclude that problematic middle bucket and likely produce even higher ROI. The 3.0x+ bucket alone has 87.5% win rate across 72 markets - that is the real edge zone.

For a replicator: apply dom 3.0x+ as the primary filter rather than dom 2x+. In practice this means: only bet on the dominant side, only when your capital allocation is 3x or more of your hedge allocation, i.e., when you have high conviction.

3. Category filter (top category = Other = all) → NOT APPLICABLE

100% of trades are classified as "Other" because the standard category keyword set does not include LoL team names. The filter returns the identical baseline set. Zero information, zero lift, zero drag.

For a real within-category filter, the relevant dimension would be LPL vs LCK vs International and Game-level vs Series-level markets. Those cannot be computed precisely from the data provided, but the top markets by P/L suggest LPL game-level markets (Game 1, Game 2, Game 3 individual winners) outperform LCK series-level markets in absolute P/L.

4. Exclude worst hours (6, 7, 8, 14 UTC) → MODEST LIFT

Removing the four worst-performing hourly buckets (6:00, 7:00, 8:00, 14:00 UTC) improves ROI from +5.62% to +6.04% (+0.42pp) while retaining $938K of the $1.21M capital base.

The improvement is genuine but small. Hours 6-8 UTC correspond to early LPL afternoon matches (14:00-16:00 CST) - these appear to be lower-quality matchups or early-round matches where the wallet's read is less sharp. Hour 14 UTC corresponds to late match hours (22:00 CST, near the end of the evening broadcast) where the wallet is perhaps less engaged.

The win rate improvement (67.7% to 71.6%) confirms these hours genuinely drag. However, the excluded hours still produced positive absolute P/L ($56,694 vs $68,114 at baseline means the excluded hours contributed +$11,420 positive P/L despite lower ROI). This is a trade-off, not a pure win.

For a replicator: skip the first match of the day (typically starting around 07:00-08:00 UTC) and the final matches after 14:00 UTC. Focus on the 09:00-13:00 UTC window where win rate exceeds 65% in every hour and the 12:00 UTC hour alone generates +$23,358.

5. Combined filter (price 30-70 + dom 2x+ + skip worst hours) → HIGHEST ROI BUT LOWEST COVERAGE

The stacked filter achieves the highest ROI (+6.99%) but cuts capital deployed to $237K (20% of baseline) and absolute P/L to $16,572 (24% of baseline). The win rate of 58.3% is puzzling - lower than the dom 2x+ alone (82.4%) - which suggests the price filter is working against the dominance filter by removing many of the high-price dominant legs while retaining the lower-accuracy mid-price entries.

This is a known filter-stacking pathology: when two filters operate on different dimensions but partially cancel each other's benefits, the combined filter underperforms both individually. The dom 2x+ filter and the price filter are effectively cross-cutting the sample in ways that destroy the dom 2x+ ROI improvement.

Do not use the combined filter. Use dom 2x+ (or preferably dom 3.0x+) as the sole filter.

---

What filters would add real value for a replicator

The standard filter battery partially misses the dimensions that matter most for this strategy. The three genuinely useful refinements:

The single most useful "filter" for a replicator is to never bet the 2.0-3.0x conviction range - those are the markets where the wallet expressed medium-strong conviction but the information read was wrong 43% of the time. Either fully commit (3x+) or keep it balanced (1.5x or under). The middle ground is where the losses live.

---

Bottom line

Apply one filter to this strategy: wait for 3.0x+ dominance conviction, bet the dominant side only. This converts the 87.5% win rate tier into your working universe and eliminates the problematic 2.0-3.0x tier that drags overall performance. Everything else in the standard battery either does nothing (category), mildly helps (hours), or harms absolute P/L while barely moving ROI (price band).

Source wallet: 0xcae693bcf9696a2ebf0a62de767719b45f354f85 Strategy: Live LoL esports directional betting with active hedge management Reference book: $1,211,928 BUY notional → +$68,127 net P/L → +5.62% ROI in 28 days

---

One-paragraph operator brief

Build a Polymarket position system focused exclusively on League of Legends game-winner markets (LPL and LCK circuits). Watch games live or via a low-latency live score feed. When a team establishes a clear in-game lead (baron/dragon control, gold differential, teamfight win), buy their "Winner" shares at market. Simultaneously carry a small opposing hedge (10-20% of clip size) to recover capital if the game reverses. As the price moves in your favor, sell the hedge position into the orderbook rather than letting it settle at zero. Cap your primary conviction clips at $1,000-$5,000 per game. Target the 09:00-13:00 UTC window (LPL primetime). Only commit large clips when you are at least 3x overweight on one side (dom 3.0x+). Expect +5-7% monthly ROI on $1M+ of cycling capital, or proportionally on smaller books.

---

1. Market Selection

Circuit priority within LoL:

LPL game-level markets generate the most absolute P/L in the reference book (NiP vs BLG Game 1: +$12,464; BLG vs WE Game 3: +$4,000; WE vs NiP Game 2: +$4,742). The LCK markets are more competitive and the T1-related markets specifically are crowd-distorted (T1 popularity creates persistent overpricing on T1 outcomes). The reference book lost -$5,799 on BNK FEARX vs T1 Game 1 and -$591 on HLE vs DK BO3.

Scheduling awareness:

The reference book's peak P/L hour is 12:00 UTC (+$23,358), corresponding to ~20:00 CST - peak LPL evening primetime. The second-best hour is 11:00 UTC (+$9,661). Plan to be at your desk or have your system active for the 09:00-14:00 UTC window on LPL broadcast days.

LPL Group Stage and Playoffs run April-June and July-August. LCK runs in parallel. This strategy has a natural off-season (December-February) when LPL and LCK are on break. The May 2026 window is an excellent operating environment.

---

2. Entry Logic

The entry trigger is an in-game event, not a price level. The wallet does not anchor to $0.60 or $0.70 - it enters wherever the orderbook offers shares when the in-game signal fires.

def should_enter(market, game_state_signal):
    # Market universe check
    if not is_lol_game_market(market.slug):
        return None
    
    # Hour check - skip early and late session
    if utc_hour(now()) not in range(9, 15):
        return None
    
    # Game state signal required
    # Signals: baron taken, teamfight won decisively, gold lead > 5k, 
    #          inhibitor down, elder dragon secured
    if game_state_signal is None:
        return None
    
    # Skip T1 markets for large clips - crowd premium risk
    if "t1" in market.slug and clip_size > 500:
        clip_size = min(clip_size, 200)
    
    # Entry: market price of the leading team
    # Accept any price - do not filter by $0.30-$0.70
    return game_state_signal.leading_team

Entry price discipline: The reference wallet buys at every price from $0.20 to $0.99 depending on the game state. The best ROI band is $0.40-$0.50 (+10.1%), which corresponds to teams that have just taken the lead but the orderbook still prices them near coin-flip. Prioritize entries in this zone when the game state clearly favors one team but the odds haven't caught up.

Probe fills: The CSV shows dozens of $3-5 fills during each game, especially early in the match. These are position-monitoring fills that let you track price movement and test orderbook depth without committing large capital. Run these probes continuously during active matches to stay oriented on the current market price.

---

3. Exit Logic (both the hedge and the conviction leg)

The exit strategy is the most complex component of this playbook. There are two separate exit paths:

Path A - Selling the hedge leg: When your primary conviction is confirmed (the team you bet on is winning decisively), the opposing hedge position becomes a recovery asset. Sell it into the orderbook aggressively rather than holding to zero.

# Hedge leg exit - sell into any bid above 0.05
if hedge_position.current_price > 0.05:
    post_sell_ask(hedge_leg, price=current_bid * 0.99)
    # Accept worst-case $0.05-$0.30 recovery on shares bought at $0.10-$0.30
    # Better than settling at $0.00

# Don't chase the last penny - if the bid is $0.02, let it settle
if hedge_position.current_price < 0.05:
    hold_to_resolution()  # Accept $0 settlement

Path B - Selling the primary conviction leg: Take partial profits when the price rallies but before the game is definitively over. The reference wallet sells into price rallies of 10-20 cents above entry.

# Primary conviction leg - stagger sell ladder
for tranche in stagger(entry_price + 0.15, 0.99, n=4):
    post_ask(primary_leg, price=tranche, shares=total_shares / 4)

# If the team actually wins (near-certain state), 
# sell the final tranche at $0.98-$0.999 rather than holding to $1.00 resolution
# The difference: certainty of $0.99 now vs gambling on resolution processing

Terminal exit: The CSV shows multiple large SELL exits at $0.999 right before resolution (e.g., 10,000 shares at $0.9990 for $9,989). These are full close-outs on winning positions. Rather than waiting for settlement at $1.00, selling at $0.999 captures the same value immediately and frees capital for the next game.

---

4. Sizing Model

Primary sizing rule: conviction-scaled clips bounded by both position size and bankroll.

The 2.0-3.0x tier is a caution zone. The reference wallet wins only 57.1% of dominant-side bets at this conviction level, well below the 87.5% of the 3.0x+ tier. Reduce sizing at this tier rather than scaling up with apparent conviction.

Bankroll scaling:

The reference wallet deployed $1.21M over 28 days at +5.62% ROI. The natural capacity ceiling for this strategy is approximately $2-5M of monthly notional before you begin moving the thin LoL markets against yourself. Above that, fragment across multiple wallets.

---

5. Both-Sides Allocation

The hedge structure is integral to this strategy - do not remove it.

Target hedge ratio: 10-20% of primary clip size on the opposing side of each market.

When to hedge:

• Always add a small opposing hedge when your primary clip exceeds $500
• The hedge protects against sudden game reversals that the orderbook reprices before you can exit
• The median second-side lag of 544 seconds (9 minutes) means the wallet typically opens the primary position, watches for 9 minutes, then adds the hedge when the game confirms the primary read

When NOT to hedge:

• Small probe fills ($3-50) - the cost of hedging outweighs the insurance
• Games with less than 3 minutes remaining (insufficient time for the hedge to be relevant)
• When the dominant team is at $0.95+ price (the hedge leg at $0.05 rarely recovers enough to justify the cost)

Hedge exit rule: Sell the hedge immediately when the primary position reaches its target sell price. Don't leave the hedge open after you've exited the primary conviction leg - you've already captured your profit and the hedge becomes a speculative underdog bet.

---

6. Hour Scheduling

Day-of-week priority:

---

7. Information Sources for Signal Generation

The edge in this strategy is the informational gap between what is happening in the game and what the Polymarket orderbook currently prices. The orderbook lags behind the game state by approximately 30-120 seconds for large moves.

Required data feeds:

Signal hierarchy (in order of reliability for position sizing):

1. Baron Nashor secured by one team: strongest signal. Baron control often determines game outcomes. Buy the baron-taking team immediately at whatever price exists.
2. Inhibitor down: the team with inhibitor advantage has structural map pressure. Medium-high signal.
3. Elder Dragon secured: late-game dominant signal. Buy the elder team heavily.
4. Gold lead > 5,000: sustained advantage. Medium signal, size at 1.5-2.0x tier.
5. First blood / first tower: early signal, weak on its own. Probe only.

---

8. Risk Management

Per-session and per-series limits:

Per game max exposure:   $5,000-$8,000 primary + $500-800 hedge
Per series max exposure: $15,000 across all games in the series  
Per day max exposure:    $50,000 total BUY notional
Per week drawdown limit: If -$5,000 net for the week, reduce to probe-only mode

The reference wallet's worst week was still positive (+$7,790 in the weakest week). A -$5,000 week would be an anomalous underperformance and should trigger a review of the conviction tier allocation.

---

9. Diagnostic Checklist

Run weekly to verify the strategy is still working:

---

10. What This Playbook Deliberately Does NOT Include

• No price-band filter. The wallet enters at every price from $0.20 to $0.99. Filtering to $0.30-$0.70 destroys absolute P/L while barely improving ROI. The game state determines fair value, not the current price level.

• No fixed-time entry. There is no "enter 5 minutes before the match" rule. Entry is triggered by in-game events. Without a live game state signal, do not enter.

• No copy-trading. Each game is a discrete event that resolves in 30-50 minutes. By the time you see another wallet's position, the game state that motivated the position has already changed.

• No scaling into the 2.0-3.0x conviction range. This is the most counterintuitive rule in the playbook. Higher apparent conviction in this tier actually correlates with lower accuracy (57.1% vs 87.5% at 3.0x+). The most likely explanation is that the 2.0-3.0x bucket represents markets where the wallet scaled into a deteriorating position - averaging down when the read was wrong. Do not reward escalating commitment.

• No T1 large clips until calibrated. T1 is the most crowd-distorted team in Korean LoL. The public overprices T1 consistently, which means the actual odds are worse than the Polymarket price suggests. The reference wallet's two biggest losses are both T1 series.

• No 24/7 bot operation. This strategy requires in-game attention. Running automated orders without live game state will produce random directional fills with no edge. The worst thing you can do is automate entries without the signal engine.

• No over-hedging above 20% of primary clip. The reference wallet's paired cost of $1.27 is already expensive insurance. Hedging at 30%+ of clip size makes the insurance unaffordable and requires even higher directional accuracy to break even.

The entire value of this strategy is concentrated in a specific type of attention: watching LoL games with enough experience to read the game state correctly and fast enough to enter before the orderbook catches up. Everything else in the playbook is implementation detail around that core competency.