Today's best ideas
1
Pick an ideaGreen BUY NOW means it passed every check. Amber WATCH means it didn't.
2
Buy with a limit orderUse the "price to pay". Skip it if the stock has already moved away.
3
Sell by the planEvery card says when: target, stop and a date. A SELL banner lights up when it's time.
More · market flow, insider buys, watchlist, charts and signals
Watchlist · tap a ticker for its chart
More signals · Trade Alert, Cheap Plays, tested setups, ideas by ticker, how to read the app
Ideas
Details · how the numbers are made
More
ScannerThe whole watchlist ranked by setup strength, with each ticker's price range and dealer levels.
Track recordHow the price model and the trade rules did in out-of-sample tests, and live results so far.
Swing & LEAPS detailEvery ticker's swing pick, LEAPS pick and the tests behind them.
Cheap Plays$5–$50 options expiring in 1–2 weeks, when the model expects them to pay.
Week planOne ticker's forecast for this week and its trade idea (uses the ticker picked at the top).
Flow by tickerCall vs put premium and net bullish / bearish skew for every ticker with activity.
How to read this app · in plain English
- Ideas is the place to start: six tabs of trade ideas. Every card shows the exact contract, the limit price to enter at, the most you can lose, the most you can make, the break-even price, the odds of profit and of reaching the target, and the expected value (EV: the average result per trade if you repeated it many times).
- Flow shows unusual option activity: big trades, what share was bought at the ask, and whether the money is leaning bullish or bearish. Side and timing are inferred from 15-minute-delayed snapshots, not from the tape.
- Chart and GEX show the price chart, the model's price ranges and the levels where option dealers hold the most exposure (walls and the flip) that tend to act like magnets or barriers.
- Odds come from a price simulation of that exact contract, not from a hunch. Where an idea type has no backtest, its card says so.
Options Picks — swing trades, LEAPS and cheap plays
Swing trades: the best call, put or debit spread per ticker for a 1–6 week hold, only when reward:risk is at least 2:1 and the expected value is positive after costs. LEAPS: long-dated calls ranked by long-term signals that were tested. Cheap Plays: $5–$50 options expiring in 1–2 weeks, only when the model expects them to pay after costs. Every number comes from the live option chain and a backtest with its sample size. Nothing here is a sure thing.
Best swing trades — top 3 by expected value per $ risked
Every ticker — tap a row for the signals behind it
Details · how swing trades are picked and graded
Price chart · daily
Details · price outlook by horizon, probability ladder, key levels, model drivers
Price Outlook — probabilistic ranges, out-of-sample tested
★ next to a price = most likely level/bucket. The star rating (1–5) shows how strong the setup is compared with this ticker's own history. Click a row to see its full probability ladder.
Options flow
GEX Heatmap — net dealer exposure by strike × expiration
Details · how the heatmap is built
Details · exposure by strike, gamma profile, expected moves, max pain, unusual activity, methodology
Net exposure by strike
Gamma profile — total net GEX if spot moved (IV held fixed)
The gamma flip (zero-gamma level) is where this curve crosses zero, re-pricing every contract's gamma at each hypothetical spot.
Setup scanner — watchlist ranked by star rating
Backtest columns are this ticker's out-of-sample 1-week numbers. "Edge" = model hit rate minus the hit rate of simply always guessing the majority direction. Edge near 0 means no real directional skill.
Dealer gamma across the watchlist
Click a row to open its GEX view. Net GEX = $ change in dealer delta per 1% move (naive convention: dealers long calls, short puts). Flip distance is the gamma flip vs spot.
Details · unusual options activity across the watchlist
Equity curve — following the signals vs buy & hold (out-of-sample, after costs)
Performance by star rating — did more stars mean better calls?
Trade-plan simulation — entry, 2×ATR stop, target 1 (1-month P75/P25), time stop
Details · accuracy by horizon and probability calibration
Accuracy by horizon
Probability calibration — 1-week P(up): predicted vs actual
Details · methodology
- Walk-forward, out-of-sample only. Models are refit every 42 bars on an expanding window. Each refit only uses rows whose future outcome was already known on the refit date (purged labels), then predicts the next block it has never seen. The first ~2 years train the model. The next 126 out-of-sample bars only set the star scale and are left out of every statistic.
- Ranges: a GARCH(1,1) + EWMA volatility forecast scaled to each horizon, multiplied by the ticker's own historical distribution of standardized moves (fat tails and skew included). That distribution is then tilted so P(up) matches the direction model.
- Direction: gradient-boosted trees + regularized logistic regression on 28 technical features (momentum, trend distance, RSI, MACD, Bollinger %B, volatility regime, volume, 52-week range). Probabilities are re-calibrated using only past out-of-sample results, so a model with no skill shrinks toward the base rate.
- Stars: where today's weighted score (1D 10% · 1W 40% · 2W 30% · 1M 20%) ranks against this ticker's own past out-of-sample scores, minus one star if the horizons disagree. 3★ or more triggers Buy/Sell. Otherwise it's Hold.
- Equity curve: the position is set at the close of day t and earns the return from t to t+1. Costs are 5 bps (equities) or 10 bps (crypto) per unit of position change. The long/short version shorts on Sell signals.
- Chart signals: BUY/SELL markers are the walk-forward signals on the day they fired; TAKE PROFIT / STOP / TIME markers are what the same trade plan (2×ATR stop, 1-month P75/P25 target, 21-bar time stop) did afterwards. Idea markers on SPY/QQQ come from the 2021–2025 out-of-sample test of the trade-idea rule. GEX level events use levels known at the previous close and are judged on the next bar.