No single signal should be able to blow up your day, week, or month. That's it. Everything below is just discipline in service of that principle.

The corollary: the dashboard's risk controls (/account/risk) are not optional. The daily loss cap, per-trade cap, and 24h kill switch exist because judgment fails at 2am when you're tilted. The rails catch you.

The full Kelly criterion says size = edge / odds. In prediction-market terms, if your edge is 10 cents on a 50¢ contract, that's 20% of the price — a 20% Kelly bet.

Don't actually bet 20% Kelly. Half-Kelly (10%) is more robust to mis-calibration, and quarter-Kelly (5%) is what most disciplined subscribers use.

Translate to dollars: if you have $10K of trading capital and quarter-Kelly says 5%, that's $500 max on the trade. The 'max entry' price in the signal tells you how many shares that buys you.

If the market is already past the caller's max entry. The edge they identified is already gone — chasing erodes you.

If you can't articulate the thesis after reading. Either the caller didn't write enough, or you don't understand the catalyst. Either way: skip.

If you've already hit your daily loss cap (or any user-imposed limit). The rail trips for a reason.

If the time horizon doesn't fit your style. A 'long' horizon signal (until-resolution months out) is not a 'jump in for an hour' trade.

If you're tilted. After two losses in a row, take a 60-minute break. The signal will still be there; your judgment won't.

Pick a number — say 2% of capital — and treat it as a hard ceiling. If you hit it before noon, you're done trading for the day. No 'one more.'

When /account/risk's daily loss cap ships server-side enforcement (phase 5), it'll do this for you automatically. Until then, write the number on a sticky note and respect it.

Why daily? Because daily caps prevent the worst pattern in trading: revenge-trading after a loss. Walking away for 24 hours is usually enough for the impulse to pass.

If you find yourself thinking 'this is the one,' that's the cue to size smaller, not larger. The brain inflates the importance of the current trade because the past is already gone and the future is uncertain.

Across hundreds of trades, position-level outcomes are noise. Process-level outcomes (Did you follow your rules? Did you size for the worst case? Did you skip when the thesis was unclear?) are signal.

Track your process, not your daily PnL. The PnL follows from the process, not the other way around.

Phase 5 lands custodial-wallet trading + the auto-copy rail. At that point, /account/risk's per-trade + daily caps become server-side hard limits — you literally cannot exceed them from any path (dashboard, Discord, Telegram, auto-copy).

Setting your caps before that ships is recommended. Defaults are conservative ($500 daily / $200 per-trade) — you can tune them, but raise them only when you've calibrated against the discipline.