HOTApr 2Vesper v1.3: Kalshi temperature markets added — scans 19 city series (NYC, Dallas, DC, Boston, Chicago, etc.), compares NOAA forecast high to threshold, fires when 3°F+ clear with 8%+ edge
NEWApr 2Vesper v1.3: signal outcome tracker — logs every signal that fires (entered or not) to vesper_signal_log.json. Tracks resolution over time to validate edge quality
NEWApr 2Vesper v1.3: calibration bias logger — tracks extreme-priced markets (85–97% YES, 3–15% YES) with $5k+ liquidity to vesper_calibration_log.json. Building favourite-longshot bias dataset
FIXApr 2Vesper v1.3: one-sided market pricing fixed — when bid=0 but ask>0, use ask directly (was computing mid=0.005 and failing 0.01 floor). Floor also lowered 0.02→0.01
FIXApr 1Vesper v1.3: Polymarket weather fetch removed — "weather" tag returned sports/politics (mislabeled). Was scanning 2,000+ irrelevant markets per cycle for nothing. Replaced entirely with Kalshi
HOTApr 1Vesper v1.3: Kalshi rain markets added — scans NYC, Houston, Miami, Seattle, New Orleans, Dallas, Austin, Denver, SF, LA, Chicago. NOAA precip probability vs market price, 8%+ edge required
HOTApr 1Seraphina v12: RSI_OB exits replaced with trail arm — RSI > 73 now arms a 0.5% trailing stop instead of exiting immediately. Trail exit has 12W/0L record vs RSI_OB's 74% loss rate
NEWApr 1Seraphina v12: ATR volatility filter — skips new entries when ATR is outside 20th–85th percentile of 75-candle window. Filters dead consolidation and flash crash conditions
FIXApr 1Seraphina v12: TP raised 3.5% → 5%, max open positions reduced 12 → 6. Reduces simultaneous SL exposure during macro moves
NEWApr 1Vesper: max open positions raised 8 → 16. Position cap was too conservative for paper trading with 13–15 signals available per cycle
FIXApr 1Bots now run under restart wrappers — auto-revives in 5s if process crashes. Previously a single exception would kill the bot until manually restarted
FIXApr 1Dashboard: Vesper toggle was stuck on STARTING... — loadBotStatus() was not updating Vesper's toggle state. Duplicate loadVesper() redefinition inside setInterval also caused JS crash
HOTMar 31Vesper v1.2: momentum edge grounded to nearness-to-mid, volume signal requires 1–5% price move window, reversion signal removed (prediction markets move toward 0/1 not 0.5), dedicated weather fetch scanning 8 tags, tag diversification cap
HOTMar 31Vesper v1.1: liquidity filter ($2k min), tiered stop-loss by liquidity, volume direction fix, position sizing by signal type, expanded weather detection (50+ keywords, 100+ cities)
FIXMar 31Seraphina v12: RSI churn bug fixed — rsi_sell_min raised 65 → 73 (was below rsi_buy_max=70, causing instant buy→exit loops). Added 20-min minimum hold gate
FIXMar 31Login: removed 400ms debounce auto-submit that fired mid-password. Now Enter-only submission
HOTMar 15Loachy v4: real de-vigged edge detection — consensus prob normalized to remove house vig, edge = true_prob − best_price_implied. No more fake line-shopping "edge"
FIXMar 15Loachy v4: Kelly bet sizing now uses live wallet.cash (was fixed $50 starting budget regardless of bankroll growth/loss)
FIXMar 15Loachy v4: overnight skip changed to 2–6 UTC (was 0–8, was cutting NFL primetime + NBA night games every session)
FIXMar 15Loachy v4: pending expiry = min(90min, game_start−30min). Was 30 min — bets expired before next 60 min scan could place them
FIXMar 15Loachy v4: weather uses real team city coordinates — was using Kansas (geographic center of US) for every single game
NEWMar 15Loachy v4: INDOOR_TEAMS set — domed stadium teams skip weather lookup entirely. MAX_LIVE_CALLS raised to 4 (was 1)
NEWMar 15Loachy v4: min_edge = 2% real de-vigged, min_book_count = 3. AI prompt redesigned as explicit veto mechanism with true_prob and edge% context
HOTMar 15Jacob v5: complete rewrite — daily candles (3mo) for MA50 + RSI14. Per-ticker MA50 gate is hard requirement for any entry
NEWMar 15Jacob v5: signal-based exits — RSI > 65, TP +2.5%, trailing stop (+2.5% arm / −1.5% fire), hard SL −6%. No more time-based 2-day exits
FIXMar 15Jacob v5: options disabled — 4% round-trip spread destroys edge at $1k budget. Long-only (SHORT direction removed)
FIXMar 15Jacob v5: removed mock fallback — Yahoo Finance failure now skips ticker (no fake data trades). Score threshold raised to 68
HOTMar 14Seraphina v9: grid trading replaced with Hybrid Trend + RSI + Funding strategy. MA50 trend filter + RSI(14) entry <35 required simultaneously
NEWMar 14Seraphina v9: Binance funding rate overlay — credits income every 8h when rate > 0.03%/8h. 2.5% TP with trailing stop
FIXMar 14Seraphina v9: fee drag resolved — grid was losing 0.62% per round-trip vs 1% grid spacing (62% of gross profit to fees)
NEWMar 12Loachy v3: approval watcher thread, CLV tracking, parlay builder, sport_key fixes, same-game protection
NEWMar 10Farm homepage: mountains, camping van, moon easter egg, live bot stats, password gate, bossa nova music