mirror of
https://github.com/jaxperro/winning-wallet-finder.git
synced 2026-08-08 05:37:46 +00:00
value: developer README + architecture-table row — silo rules, ops, verdict criteria
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
jaxperro
@@ -26,6 +26,7 @@ Three deployed pieces + one static dashboard:
|
||||
| **daily pipeline** (`live/daily.sh`) | this Mac, launchd **08:00** (a `pmset repeat wakeorpoweron` RTC wake at **07:58** makes this run on time even from sleep — set 2026-07-08, undo with `sudo pmset repeat cancel`; job wrapped in `caffeinate -i` so mid-run idle-sleep can't kill it. Heads-up: the wake fires even lid-closed-in-a-bag — cancel it if the Mac travels) | refresh the bet cache → 5-gate skill scan → fee-aware sharp selection → conviction floors → backtest book → publish JSON feeds to GitHub |
|
||||
| **copybot worker** (`copybot.py` via `host/start.sh`) | **Fly.io app `wwf-copybot`, region `arn` (Stockholm), 24/7** — migrated off Railway 2026-07-06: Railway ran it in a US region, which Polymarket's IP geoblock would 403 the moment orders got real; Stockholm is unrestricted AND ~25ms from the CLOB's eu-west-2 primaries. Every boot self-checks the geo-gate (`host/geocheck.py`) and verifies the clone against `git ls-remote` (stale-replica guard) | **T0 = the RTDS trade stream** (gotcha 18): every platform trade, wallet-attributed, **~1s detection** (measured p50 0.8s over 22k msgs), silent-stale guard force-reconnects a quiet socket. Backstops: the Alchemy address-activity webhook (~3s, `POST /alchemy`, signature-verified — **T0b chain seed** 2026-07-14: its tx hashes decode straight from the receipt's OrderFilled logs when the data-api indexer lags, so RTDS-silent markets still copy at push speed), a 5-min backstop poll, and the reconcile janitor. Per-wallet **trade cursor** paginates the activity feed so clip bursts can't scroll past a fetch (H3). 60s heartbeat settles/publishes; paper-copies with real fees/lag/slippage AND live-FAK thin-book reality (**paper FAK parity** 2026-07-16: a quote with no ask inside the protected band books a MISS, not a fill — keeps the live-vs-paper per-signal ratio honest; `live/edge.py` turns it into the daily bankroll-decision verdict in `history/edge.csv` + the Discord digest footer); commits its book back to the repo |
|
||||
| **REAL-MONEY worker** (same code, `COPYBOT_ROLE=live`) | **Fly.io app `wwf-copybot-live`, `arn`, 24/7, ARMED 2026-07-10** — separate app so the two books never share a process, state file, or failure mode. Trades the signer's **Deposit Wallet** `0x455e…45a1` (pUSD collateral) via the unified SDK (`polymarket-client`) | Same T0 RTDS detection (`RTDS_DETECT=1`) + Alchemy/poll backstops; **own-fill push** via the CLOB user channel (`userws` in the heartbeat) makes in-play pending holds adopt the moment they match — ws events only *trigger* the resolver, `get_order` stays the arbiter (gotcha 17). Paper-parity sizing (4% of equity, $1 venue floor, no price floor, hard caps retired — the **depth gate** is the scaling rail: skip spread>0.08 or ask5c<$50, stake ≤10% of 5c depth). Every placement/exit/settle pings Discord. Disarm: `flyctl secrets unset LIVE_CONFIRM -a wwf-copybot-live` |
|
||||
| **VALUE bot** (separate experiment, [`value/README.md`](value/README.md)) | **Fly.io app `wwf-valuebot`, `arn`, 24/7, PAPER** — a hard SILO: own code (`value/valuebot.py`, zero copybot imports), own state/feed, own image, no shared wallet ever | Systematically buys sub-2¢ contracts (the calibration study's one underpriced bucket — [`value/PLAN.md`](value/PLAN.md)); honest FAK fill model, chain-truth settles, $1 tickets, event cap 1. Verdict at ~2k resolved tickets vs a ~1.05% break-even hit rate |
|
||||
| **Discord digest** (`live/discord_daily.py`) | end of the daily pipeline | one message/day: the sharp list with profile links + 30-day conviction stats (per-trade pings retired 2026-07-04 for PAPER; the LIVE book pings every real placement/exit/settle) |
|
||||
| **dashboard** | [jaxperro.com/trading](https://jaxperro.com/trading) + [jaxperro.com/live](https://jaxperro.com/live) (static, in the `jaxperro` repo) | `/trading` renders the paper book, backtest book and sharp table; `/live` is the REAL MONEY page (reads `live/copybot_live_real.json` — live since 2026-07-10). Both pages share `trading/copybot-section.js` — one renderer, no drift |
|
||||
|
||||
|
||||
@@ -0,0 +1,84 @@
|
||||
# VALUE — the undervalued-market paper bot
|
||||
|
||||
A **silo'd** experiment: can systematically buying sub-2¢ Polymarket contracts
|
||||
beat their price? Research says yes on paper (~1.24x gross); this bot exists
|
||||
to find out if the fills are real. It shares **no code path, state, app, or
|
||||
wallet** with the copy trader — see *Silo rules* before touching anything.
|
||||
|
||||
## Why this exists (30-second version)
|
||||
|
||||
A calibration study over 13.5M trusted resolved bets (2026-07-17, details in
|
||||
[PLAN.md](PLAN.md)) found Polymarket well-calibrated above 60¢, longshots
|
||||
2–60¢ systematically OVERpriced — and the tail **below 2¢ UNDERpriced**:
|
||||
entries at an average 0.82¢ resolved winners 1.02% of the time (1.24x, ≈+20%/$
|
||||
net of fees, z≈+12 market-clustered). The effect survives excluding skilled
|
||||
wallets and refunds, and concentrates near expiry. History can't prove the
|
||||
asks are actually fillable — that's this bot's one job.
|
||||
|
||||
## How it works (valuebot.py, single file, pure stdlib)
|
||||
|
||||
Every 5 minutes:
|
||||
1. **Scan** gamma for active markets, soonest-ending first (the edge lives
|
||||
near expiry; gamma 422s past offset ~2000, which conveniently caps the
|
||||
scan to exactly that window). Any outcome marked ≤ 2¢ is a candidate.
|
||||
2. **Fill or honest miss**: fetch the REAL CLOB ask ladder; a $1 FAK either
|
||||
fully fills inside `min(2¢, best_ask × 1.05)` — the same protected band
|
||||
the live copy executor sends — or logs a MISS. No phantom fills, ever
|
||||
(the 2026-07-16 paper-parity lesson, inherited from birth).
|
||||
3. **Guards**: flat $1/ticket (venue minimum = reality), event cap 1
|
||||
(correlated dust resolves together), 300-ticket portfolio cap, 6h
|
||||
re-check cooldown per token, entry taker fee by category.
|
||||
4. **Settle at chain truth**: payout vectors from the CTF contract
|
||||
(`payoutNumerators`) — wins pay 1.0, **0.5 refunds are real**, losses 0.
|
||||
Never trusts CLOB winner flags or price marks.
|
||||
5. **Publish**: state + feed + fills committed back to this repo
|
||||
(`valuebot: paper feed [skip ci]`), same pull-rebase-push discipline as
|
||||
the copy books.
|
||||
|
||||
## The verdict (what decides paper → real money)
|
||||
|
||||
Break-even hit rate ≈ 1.05% at 1¢ entries. Needs ~2,000 resolved tickets
|
||||
(2–6 weeks). Pre-registered kill criteria (PLAN.md): fill rate <30%, or
|
||||
realized multiple <1.1x after 2k tickets, or can't deploy >$50/week. Track:
|
||||
`stats` + `realized_multiple` + `fill_rate` in the feed. Early losses are
|
||||
EXPECTED — at ~1% hit rate the first hundreds of tickets are mostly Ls; the
|
||||
multiple is meaningless until refunds/wins accumulate.
|
||||
|
||||
## File map
|
||||
|
||||
| file | what |
|
||||
|---|---|
|
||||
| `valuebot.py` | the whole bot — scan, fill model, settle, publish |
|
||||
| `PLAN.md` | research findings, strategy spec, rollout gates |
|
||||
| `valuebot_state.json` | the book (bot-committed; don't hand-edit while it runs) |
|
||||
| `valuebot.json` | published feed (future jaxperro.com/value reads this) |
|
||||
| `valuebot_fills.jsonl` | append-only fill ledger |
|
||||
| `start.sh` | Fly boot: clone main, ls-remote clone-guard, exec bot |
|
||||
| `fly.Dockerfile` | python:3.12-slim + git; NO pip deps by design |
|
||||
| `../fly.value.toml` | app `wwf-valuebot` (arn); **no http_service** on purpose — nothing auto-restarts it under a state edit (README gotcha 15) |
|
||||
| `../tests/test_valuebot.py` | 6 stub tests — fill band, thin-miss, event cap, cooldown, settle math |
|
||||
|
||||
## Ops
|
||||
|
||||
```bash
|
||||
python3 tests/test_valuebot.py # no-network stub tests
|
||||
python3 value/valuebot.py --once # one real scan cycle, no publishing
|
||||
flyctl logs -a wwf-valuebot --no-tail # watch cycles ("book: cash …")
|
||||
flyctl deploy --remote-only -c fly.value.toml --ha=false -a wwf-valuebot
|
||||
# image change (start.sh/Dockerfile);
|
||||
# code changes need only a machine restart
|
||||
```
|
||||
|
||||
Secrets on `wwf-valuebot`: `ALCHEMY_RPC_URL` (chain settles),
|
||||
`GITHUB_TOKEN` (fine-grained PAT, contents-RW this repo — state durability +
|
||||
feed). Without the token the bot trades but rewinds to the last committed
|
||||
state on restart (it warns at boot).
|
||||
|
||||
## Silo rules (do not break these)
|
||||
|
||||
1. Nothing under `value/` may import `copybot.py` or `copytrade.py` — the
|
||||
~60 lines of book/fee/payout helpers are duplicated here ON PURPOSE.
|
||||
Read-only research libs (`trust`, `smart_money`, `payouts`) are allowed.
|
||||
2. Never share: state files, feeds, Fly apps, webhooks, or (when real money
|
||||
ever comes) the wallet. The copy trader must be un-affectable from here.
|
||||
3. `bankroll`/stake/cap changes and the paper→real gate are the USER's call.
|
||||
Reference in New Issue
Block a user