Today's picks.
▸ Showing the last session — Thu Jun 18. New picks fire after ~9:30 ET.
Live recommendations from validated plays. BUYs are deterministic-gate-cleared, then LLM-scored for conviction tier (HIGH/MED/LOW). SKIPs are grouped by the override mechanism that filtered them. All judgment cites the validated mechanism — no caution-as-rigor.
New confirmed BUYs today (0)
Genuinely new picks that made it through all four
stages today: mechanical gate passes, advisor evaluates favorably, AND
the play's publication policy is satisfied on fire #1 (plays like
csp_panic and leaps_drip that publish on the
first fire by design — single-fire WR is real edge, see methodology).
Separately, 1
re-confirmation
fired today on ticker
already in the open book — those are signal-validation only (do not
add to existing positions) and are listed below, not counted here.
Position re-confirmations (1) — signal validation only, do not add
These tickers are already in the open book and the play's edge re-fired today. They confirm the existing thesis — keep holding per the original exit rule. Not a new entry: do not add to the position (for option-sellers, the existing contract is still the trade).
Open positions (10) · ranked by validation tier
One row per ticker, not per fire. Re-fires within each play's hold window are
treated as signal confirmations on the same position (per spec.position).
Sorted by freshness.
Click a column header to re-sort. Entry recomputed from raw bars.
Signal freshness: Fresh 0
Active 4
Mature 6
— factual labels only; no claim about whether to chase or hold (per
backtest 2026-05-26, neither chase nor mean-reversion edge clears permutation per play).
Sectors: ★ AI/DC 1 — AI/DC closed n=142 mean +15.9% WR 68% p=0.0000. Software closed n=38 mean +6.4% but median -3.6%, WR 45%, p=0.75 (mixed). Per validation 2026-05-26. Labeling only.
Two problems. First, validation: this fired from source 'short_covering', which has no validated cell or spec evidence in the lookup (label not_tested) — there is no backtest edge to anchor on, unlike the RSI≤≅15 short_covering_rsi15 base. Second, the setup is broken, not oversold: RSI 11.3 looks like an extreme-reversal trigger, but it sits on a -66% five-day collapse with the stock at the 3rd percentile of its 52-week range. RSI mean-reversion is for names that overshot, not names that are imploding on a structural event — same failure mode as VRDN. Elevated option RVOL (4.3) is just as consistent with a scramble as a bottom. I can't responsibly source a catalyst for a turn from our data, and the -66% move is extreme enough to warrant a data-quality check on the bars. Single un-validated thread on a broken chart.
institutional_oi_build's backtest is the validation and it holds — score>=9 onset OOS PF 3.54 / WR 69% on n=80, perm_p=0.0046, regime-robust — so the play isn't the problem; this specific BORR instance is. The headline "3 distinct plays on the ticker" is illusory stack depth: uoa_smart_divergence (call_pct 0.874, vol_ratio 7.66) and institutional_oi_build (call_pct_avg 0.879, rvol 4.45, total_oi 36K) are the SAME LEAPS-call accumulation read on the SAME chain — one independent thread, not two. And the third "play," institutional_oi_trap_fade, isn't corroboration at all: it's a SHORT, and it fired on BORR 2026-05-29 precisely because this build is failing — both trap legs tripped (price fwd_5d −6.18% AND LEAPS call OI declining −3.27% four days post-signal, derived straight off build_pick 2026-05-25_BORR). The system's own machinery is now betting against the long. Two quality marks compound it: the build scores exactly 10/17 (at the floor — fine, that's the peak-alpha band) but leaps_pct_avg is 0.739, which lands in the 70%+ LEAPS bucket the spec flags as WEAKEST (PF 1.42, leaps_pts earned = 0); the validated edge lives in the 30-50% bucket, not here. The price structure seals it — this is a flow play so "no catalyst" is by design, but the technical setup is broken, not intact: BORR sliced through its multi-month ~$5.50 shelf on a 30.6M-share capitulation candle (5/21, vs ~7M typical) and kept bleeding to $5.07, now −23% off the 60-day high. A single options-chain thread, fired into a failed-support breakdown, that the platform's own trap-fade is already shorting. Single-thread bet the surrounding context refuses to confirm.
Provisional (56) — single fires from playbook-eligible plays, watching for confirmation
Single fires of second_fire plays — preliminary signals, not yet
held. A confirming second fire on the same ticker (any play) promotes them to
Open. If none arrives within the hold window, the signal closes to the Track
Record. See docs/POSITION_MODEL.md.
Options — cash-secured puts (2)
Sell-to-open puts (csp_panic) — a velocity play: collect
premium, capture ~25% in ~2–5 days, then close. It does not hold
to expiry. Premium shown is the entry mid; the realized return lands in the track record
when the position exits. A different trade than an equity buy — columns reflect the option.
Track record
Closed positions only — one row per ticker per continuous holding period. Re-fires within 20 days are rolled into the same position. Entry / exit prices recomputed from raw market bars.
The methodology's edge is established by out-of-sample backtest, per play — there is no single blended figure. Live forward (below) is a sanity check: the published track record is still thin (since 2026-05-08).
Show all 49 closed positions (filterable)
Sold puts exit at ~25% premium capture in ~5 days (velocity), not held to expiry — so return is on capital (premium captured ÷ strike), not the stock move. The validated edge is the backtest above (csp_panic); realized per-trade P&L shows where the entry premium was recorded.
Recently exited (10)
Positions the system closed mechanically within the last 5 trading days. Final move is realized at the exit close. Cradle-to-grave audit trail — every BUY broadcast has a symmetric CLOSE.
10 recently exited positions
Get picks by email
Every published pick, as it lands — conviction score and the model's full reasoning. No spam, unsubscribe anytime.