ZonSource Option Trader

A proposed BUY CE/PE, a score above the threshold or a CANDIDATE grade is not permission to buy. ENTRY READY appears only while every strategy, validity, quote, execution and risk check the engine itself runs is passing, and only for as long as the quote behind it is fresh. Entries only: HOLD/EXIT for open positions is under Open positions.

Only orders this engine sent. In PAPER they are simulated (broker id SIM…); trades you place yourself at the broker are never shown here. Bought and sold are filled quantities.

Most of these never became orders; the rejected ones explain the engine. The scoring weights behind these numbers are unvalidated — the specification's starting values, never fitted to real data. Read them to build a view, not to place a trade.

Directional accuracy by confidence — ADVISORY

Calibration — SHADOW

Recent decisions

1. Only expiry days count. NIFTY expires weekly (Tuesdays), so this trades about once a week. On other days the status says Not an expiry day.
2. Watch the primary book (marked ★). The others exist to check that the backtest's choice still looks right on real prices.
3. Judge it only after 30+ expiries (about 7 months). A few weeks can look great or terrible by luck; the backtest's worst single day was about −₹4.8k per lot.

1. Is it running? The first tile should say Running between 09:30 and 15:05 IST, and Last check should be under 6 minutes old. Not running means the trader is stopped or was not restarted. No data on a weekday means a market holiday or a broker login problem.
2. What did it think? The chart shows every check today: one bar per 5 minutes, height = chance of a big move. Bars that reach the dashed line are picks. Usually only 0–3 bars a day reach it.
3. Was it right? 30 minutes after each check the Big move? column fills in with Yes or No. The model is useful only if its picks see big moves much more often than the other bars — compare the two rates under Was it right?.
4. Would it have made money? Look at Paper money and the paper straddles table. One day means nothing: judge it only after 30+ trades (better ~100, which takes months).

Each bar is one 5-minute check (hover for details). Filled bars reached the threshold (dashed line). Below each bar: ● a big move followed, ○ it did not, nothing = not known yet (30 minutes have not passed).

▲ pushed the chance up, ▼ pulled it down — the two inputs that moved each score most. Needed is the 30-minute move India VIX implied.

Walk-forward, last 25% of days untouched until the end, pass rule set before running. AUC 0.5 = no better than chance. Neither model beat the bot on its own signals.

Waiting for the first refresh.