🔗 Mappings 0
💰 Value bets 0
⚖️ Arbs 0
📜 Arbs history 0
🎯 Paper-arbs 0
📊 Calibration 0
📡 Live feed 0
🔗 Cross-provider mappings — live odds side-by-side
0
| Sport | Competition | Match | Market | Variation | BF Yes | BF No | PM Yes | PM No | BF time | PM time | Spread | Copy |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Waiting for cross-provider mappings — needs Betfair events flowing. | ||||||||||||
💰 Value bets
stake=$? gas=$? comm=?%
0
| Competition | Match | Outcome | Betfair | Polymarket | no-vig p | EV Betfair | EV Polymarket | PM size | Conf | Recommend | Copy |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Waiting for matched cross-provider markets… | |||||||||||
⚖️ Cross-provider arbs — back BF + buy PM-NO (or vice-versa) to lock profit regardless of outcome
0
| Match | BF leg | PM leg | BF eff | PM eff | p | Stake | Profit | Edge | Age | Copy |
|---|---|---|---|---|---|---|---|---|---|---|
| Waiting for cross-provider arbs… | ||||||||||
📜 Arbs history — every arb that flickered, with peak edge and lifetime
0
| When | Match | BF leg | PM leg | Peak edge | Peak profit | Lifetime | BF eff | PM eff | Side | Copy |
|---|---|---|---|---|---|---|---|---|---|---|
| No arb history yet — wait for live arbs to appear and disappear. | ||||||||||
🎯 Paper-arbs — Dutched two-leg paper trades
—
Every cross-provider arb the engine detects gets paper-placed
at first sighting (sub-100ms hot path). The settler resolves
outcomes per-leg (BF runner status + PM YES resolution).
Realized profit should ≈ predicted profit if the arb math is
right and both legs would have filled at snapshot prices.
No paper-arb data yet — first arb the engine detects will land here within seconds of capture.
How to read this:
- Predicted vs realized profit — should converge as arbs settle. If realized consistently undershoots predicted, partial fills or stale-ladder pricing are eroding the edge.
- Slippage gap — predicted minus realized; smaller (or negative) is healthier. Negative would mean reality beat the model — usually a math error in the engine.
- Math errors — rows where predicted profit was positive but realized came out negative by >5%. Indicates the Dutched math didn't hold at settlement (could be partial fills, runner removals, or genuine engine bugs).
v1 caveat: assumes both legs fill at engine snapshot prices and includes sub-300ms flickers. Real execution would miss many of these.
📊 Calibration — paper-trading validation (no real money)
—
Loading…
🤖 Paper-mode bot — slippage diagnostic
Re-verifies every signal at decision time against a fresh order book and
logs a place/skip decision. Compares signal-time EV (what the
calibration above claims) with realised paper P&L
(what fills would have produced). The gap = capturable-edge slippage.
No paper data yet — bot starts ~30s after service boot.
How to read this:
- Headline win rate — is the system winning more than 50% of its bets? Useful at sample size ≥ 100.
- Calibration — predicted vs realized win-rate per probability bucket. If the deltas hover near 0 across all buckets, the no-vig math is honest. A persistent +20% means the system underestimates its own picks (lucky), −20% means it overestimates (don't deploy auto-execution yet).
- P&L by EV bucket — does a "5–10% edge" actually return 5–10% on average? If not, the displayed edges are bigger than reality.
- P&L by confidence — high-confidence rows should outperform low. If 90+ scores lose money but <50 wins, the confidence formula is over-rewarding the wrong components.
Sample sizes under 100 are too noisy to draw conclusions. Wait for settled count to grow.
📡 Live odds feed
0
| Time | Provider | Sport | Competition | Match | Market | Variation | Prev | Odds | Shares | Δ / Total | Copy |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Waiting for live odds activity… | |||||||||||