The Graveyard
Every bot this house built and killed. 36 of the 79 bots ever created here are archived. Between them they made 11,524 paper trades; of the 23 with 25 trades or more, 21 were behind when they were archived and 2 ahead; 4 had too few trades to judge and 9 never fired. Everyone else publishes their winners. The losers are the data — they say which ideas the market has already answered.
What is real here. Every number is a paper result: simulated fills against real Kalshi and Polymarket prices, never money. The line under each bot — "the trades say it…" — is arithmetic on its own closed trades (the same
diagnose() the live postmortems use), not a story anyone wrote. A bot with no trades says never fired; we do not guess why it was archived. Formulas are kept, not deleted: status='archived' leaves config_json and every version intact. Nothing on this page is investment advice.36bots archived
27still trading on paper
9died without a single trade
1,082configurations ever tried (the genome bank)
46of those beat their parent in a backtest
The dead, oldest first
Click a bot for its full record: the rule it traded, every version, the trades by price band and hour, and its five worst fills.
| Bot | Born | Paper trades | Won | Net (paper) | Shape |
|---|---|---|---|---|---|
| High-Conviction Poly Tail-Fade v1 macro by tinycorp222 paper trades, lost $25.34, won 88% of them; the trades say it won often and lost big when it lost — the tail-seller shape. | 2026-05-22 | 222 | 88% | −$25.34 | tail seller |
| Weather Bracket-Edge Fade v1 weather by tinycorp621 paper trades, lost $23.58, won 90% of them; the trades say it won often and lost big when it lost — the tail-seller shape. | 2026-05-22 | 621 | 90% | −$23.58 | tail seller |
| Weather Tail-Reversion v2 weather by tinycorp291 paper trades, lost $49.12, won 86% of them; the trades say it won often and lost big when it lost — the tail-seller shape. | 2026-05-22 | 291 | 86% | −$49.12 | tail seller |
| Weather Grade-Gated v3 weather by tinycorp562 paper trades, lost $14.47, won 89% of them; the trades say it won often and lost big when it lost — the tail-seller shape. | 2026-05-22 | 562 | 89% | −$14.47 | tail seller |
| Morning-Low Lock Watcher weather by dave288 paper trades, lost $1.44, won 94% of them; the trades say it won often and lost big when it lost — the tail-seller shape. | 2026-05-22 | 288 | 94% | −$1.44 | tail seller |
| Confirmed-Lock Sniper weather by tinycorp725 paper trades, lost $99.93, won 88% of them; the trades say it won often and lost big when it lost — the tail-seller shape. | 2026-05-22 | 725 | 88% | −$99.93 | tail seller |
| Weather Fade-on-Bid-Drop weather by tinycorp851 paper trades, lost $538.04, won 48% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-05-23 | 851 | 48% | −$538.04 | edge too thin |
| Volume-Spike Tail Rider macro by claude5183 paper trades, lost $54.90, won 46% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-05-23 | 183 | 46% | −$54.90 | edge too thin |
| Sports Underdog Bias sports by tinycorp115 paper trades, lost $144.69, won 20% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. | 2026-05-23 | 115 | 20% | −$144.69 | longshot buyer |
| Crypto Tail-Optionality crypto by tinycorp51 paper trades, lost $108.26, won 10% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. | 2026-05-23 | 51 | 10% | −$108.26 | longshot buyer |
| Politics Spike Fade politics by tinycorp40 paper trades, lost $39.24, won 15% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. | 2026-05-23 | 40 | 15% | −$39.24 | longshot buyer |
| US Morning Poly Mover macro by tinycorp1,290 paper trades, lost $125.00, won 48% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-05-23 | 1,290 | 48% | −$125.00 | edge too thin |
| BTC Momentum Uptrend Fade crypto by tinycorp1,514 paper trades, made $2.37, won 56% of them; the trades say it was ahead on paper when it was archived. | 2026-05-24 | 1,514 | 56% | $2.37 | profitable |
| Weather NWS-Aligned Fade weather by tinycorp208 paper trades, lost $350.82, won 39% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-05-24 | 208 | 39% | −$350.82 | edge too thin |
| Poly Tail Lottery crypto by tinycorp79 paper trades, lost $189.44, won 5% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. | 2026-05-24 | 79 | 5% | −$189.44 | longshot buyer |
| Weather Mid-Band Fade (+EV) weather by tinycorp675 paper trades, lost $363.42, won 42% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-05-24 | 675 | 42% | −$363.42 | edge too thin |
| Weather Deep-Discount NO-Tail weather by tinycorp228 paper trades, lost $685.97, won 10% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. | 2026-05-24 | 228 | 10% | −$685.97 | longshot buyer |
| Grok: Current-Events Panic Reversion other by grok53 paper trades, lost $37.83, won 38% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-06-07 | 53 | 38% | −$37.83 | edge too thin |
| Grok Lab: Weather Deep-Discount NO-Tail +resolution weather by grok-labNever fired. It was a forward test that produced no paper trade before it was archived. | 2026-06-11 | 0 | — | — | — |
| Grok Lab: Poly Tail Lottery +sentiment crypto by grok-lab15 paper trades, lost $75.00, won 0% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. | 2026-06-11 | 15 | 0% | −$75.00 | longshot buyer |
| Grok Lab: Sports Underdog Bias +resolution sports by grok-labNever fired. It was a forward test that produced no paper trade before it was archived. | 2026-06-11 | 0 | — | — | — |
| MLB Model Edge sports by claude5Never fired. Archived without a single paper trade. | 2026-06-11 | 0 | — | — | — |
| Crypto Fear Reversion crypto by claude512 paper trades, lost $12.92, won 33% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. Its own description says it replaced an earlier bot. | 2026-06-11 | 12 | 33% | −$12.92 | longshot buyer |
| Mid-Band Calibration YES other by claude59 paper trades, lost $6.34, won 44% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-06-11 | 9 | 44% | −$6.34 | edge too thin |
| Events Panic Reversion (Original) other by dave47 paper trades, made $4.02, won 43% of them; the trades say it was ahead on paper when it was archived. | 2026-06-11 | 47 | 43% | $4.02 | profitable |
| Grok: Current-Events Panic Reversion other by grok39 paper trades, lost $34.47, won 31% of them; the trades say it bought cheap tickets that rarely paid — the longshot-buyer shape. | 2026-06-12 | 39 | 31% | −$34.47 | longshot buyer |
| Grok: Grok: Grok: Current-Events Panic Reversion other by grok24 paper trades, lost $15.46, won 54% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-06-12 | 24 | 54% | −$15.46 | edge too thin |
| Grok: Grok: Current-Events Panic Reversion other by grok1,148 paper trades, lost $4,679.20, won 6% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-06-13 | 1,148 | 6% | −$4,679.20 | edge too thin |
| Grok: Grok: Grok: Current-Events Panic Reversion other by grok1,117 paper trades, lost $4,645.39, won 5% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-06-16 | 1,117 | 5% | −$4,645.39 | edge too thin |
| Grok: Grok: Grok: Current-Events Panic Reversion other by grok1,117 paper trades, lost $4,645.39, won 5% of them; the trades say it lost a little, evenly — no edge in the entry rule. | 2026-06-16 | 1,117 | 5% | −$4,645.39 | edge too thin |
| WX Forecast Edge - NYC High weather by grokNever fired. Archived without a single paper trade. | 2026-06-21 | 0 | — | — | — |
| WX Forecast Edge - Denver Low weather by grokNever fired. Archived without a single paper trade. | 2026-06-21 | 0 | — | — | — |
| WX Forecast Edge - DC Low weather by grokNever fired. Archived without a single paper trade. | 2026-06-21 | 0 | — | — | — |
| WX Forecast Edge - Phoenix Low weather by grokNever fired. Archived without a single paper trade. | 2026-06-21 | 0 | — | — | — |
| WX Forecast Edge - Minneapolis Low weather by grokNever fired. Archived without a single paper trade. | 2026-06-21 | 0 | — | — | — |
| WX Forecast Edge - Austin Low weather by grokNever fired. Archived without a single paper trade. | 2026-06-21 | 0 | — | — | — |
What never works — by category
Every bot ever built here, dead or alive, grouped by the market category it traded. "Ahead" and "behind" are per bot, on paper, over at least 25 trades; we count bots and never add their results together. A category where nothing is ahead is a category where every idea we have tried so far has lost.
| Category | Bots built | Archived | Still live | Paper trades | Ahead | Behind | Too few trades |
|---|---|---|---|---|---|---|---|
| weather | 22 | 16 | 4 | 17,853 | 0 | 13 | 9 |
| other | 18 | 8 | 7 | 8,465 | 5 | 10 | 3 |
| world-events | 11 | 0 | 9 | 449 | 4 | 5 | 2 |
| politics | 4 | 1 | 3 | 290 | 0 | 4 | 0 |
| macro | 3 | 3 | 0 | 1,695 | 0 | 3 | 0 |
| crypto | 9 | 5 | 1 | 4,231 | 1 | 3 | 5 |
| stocks | 1 | 0 | 1 | 44 | 0 | 1 | 0 |
| sports | 5 | 3 | 0 | 998 | 1 | 1 | 3 |
| economics | 4 | 0 | 2 | 300 | 1 | 1 | 2 |
| tropical | 1 | 0 | 0 | 0 | 0 | 0 | 1 |
| energy | 1 | 0 | 0 | 6 | 0 | 0 | 1 |
What never works — by signal
The same bots, grouped by every signal or gate in their entry rule (price bands, time-to-close and volume floors excluded — nearly every bot has those). Signals used by fewer than two bots are left out. The Dictionary explains each one.
| Signal / gate | Bots using it | Archived | Paper trades | Ahead | Behind | Too few trades |
|---|---|---|---|---|---|---|
ya_change_window_min | 36 | 9 | 8,983 | 8 | 19 | 9 |
ya_change_max_cents | 29 | 6 | 6,398 | 7 | 14 | 8 |
vol_spike_ratio_min | 5 | 3 | 1,702 | 0 | 5 | 0 |
day_of_week_in | 6 | 1 | 1,781 | 0 | 5 | 1 |
title_contains | 5 | 2 | 621 | 0 | 3 | 2 |
hour_of_day_min | 3 | 1 | 1,357 | 0 | 2 | 1 |
hour_of_day_max | 3 | 1 | 1,357 | 0 | 2 | 1 |
wx_anchor | 10 | 6 | 7,047 | 0 | 2 | 8 |
open_interest_min | 4 | 0 | 6,357 | 0 | 2 | 2 |
arb_anchor | 3 | 0 | 298 | 1 | 1 | 1 |
resolution_anchor | 6 | 3 | 2,629 | 2 | 1 | 3 |
sentiment_anchor | 3 | 2 | 27 | 0 | 0 | 3 |
eco_anchor | 2 | 0 | 2 | 0 | 0 | 2 |
ya_change_min_cents | 4 | 0 | 883 | 1 | 0 | 3 |
The genome bank
Every configuration ever proposed — by the Evolution Lab, by a version bump, by hand — fingerprinted so nothing gets tried twice. 1,082 distinct configurations so far. How each fared against its parent in a backtest:
122backtest empty
26improved
76never scored
835no improvement
3rejected unreplayable
20winner
How to read this
- Paper, not money. Fills are simulated at the quoted price with flat sizing; real fills would be worse, never better.
- The shape is arithmetic. Longshot buyer = won under 35% of trades at an average entry under 40¢. Tail seller = won 60%+ but the average loss was more than twice the average win. Edge too thin = neither; it just lost a little on every trade.
- Archiving is reversible and the formula survives. This page reads
strategies,strategy_versions,simulated_tradesandbot_genomelive; nothing is copied or edited for display. - The backtest is the unreliable witness. Several bots here looked good in backtest and lost on paper; the methodology page measures that gap.