ψ Psi Quant Research

Signal Report · US

ADI HOLD

Composite decision report grounded in 1-XH strategy signals and multi-agent debate, generated by DeepSeek at 2026-08-11 09:05.

🎯 Final Decision

Action
HOLD
Weight
0%
Confidence
55%
Variant
xiaohan_gogo_v2

Reasons

Bull Case (Analyst)

- Technical signal is a clean buy (score 0.60) with a stacked bottoming structure: 2h/3h/4h bottoms all confirmed, and the 1h blue upper band is above the yellow upper band — a momentum-recovery pattern that historically precedes mean-reversion upside.
- The cross-sectional factor view is supportive at the margin: composite z = +0.27 (rank 69/252, top 27.4%) is positive in a BULL regime catalog with zero direction-flips, meaning the stock is not fighting the tape.
- The dominant factor lift is xh_gogo_score (+2.12 z) — the same family as the technical signal — so the technical and cross-sectional views are corroborating, not contradicting. Momentum_6m (+0.37) and beta_60d (+0.58) add mild tailwinds in a risk-on regime.
- News flow shows a strong recent tape: ADI advanced +3.34% on Aug 7 and +3.26% on Aug 7 (two separate reports), outperforming the S&P 500 on a strong day — evidence of active buying interest in the high-performance analog semiconductor space.
- The stock closed at $383.80, above both fast EMA (382.8) and slow EMA (380.4), confirming the short-term trend is now up after the bottoming sequence.

Bear Case (Analyst)

- The cross-sectional rank (69/252) is only mid-pack — ADI is NOT in the production basket and missed the top decile. The composite z of +0.27 is barely positive; this is not a high-conviction long from the factor model's perspective.
- The three largest factor drags are all quality/participation metrics: sharpe_60d (-0.91), volume_momentum_60d (-0.81), and close_to_high_60d (-0.76). The stock has weak risk-adjusted returns, fading volume momentum, and is far from its highs — the rally is not being confirmed by volume or relative strength.
- Insider selling is a clear red flag: Director Ray Stata sold at $400.83 and SVP Martin Cotter sold 17,594 shares at $418.00 (June 2). Both sales were above the current price of $383.80, meaning insiders were distributing into strength at levels the stock has since failed to hold.
- The stock is down ~8% from the June insider-sale levels, and the most recent session (Aug 10) showed a -1.54% drop that lagged the S&P 500's -0.06% — the bounce is already stalling relative to the market.
- Earnings risk is imminent: Q3 FY2026 results are scheduled for Aug 19. With weak volume momentum and insider selling ahead of the print, the risk/reward is skewed to the downside if guidance disappoints.

Devil's Advocate

COUNTER TO BULL:
- The "stacked bottoming structure" is a lagging artifact: 2h/3h/4h bottoms all confirmed means price already rallied into the signal — you're buying after the move, not before. The 1h upper-band crossover is a momentum-chasing trigger that fails frequently in rangebound tape.
- xh_gogo_score at +2.12 z is a proprietary, likely overfit factor; its correlation with the technical signal may be tautological (same underlying price path), not independent corroboration. If the factor is just a smoothed version of price, you've counted the same evidence twice.
- The Aug 7 "two separate reports" of +3.34% and +3.26% are the same session duplicated — sloppy data handling. If you can't verify the tape, the "active buying interest" claim is unsubstantiated.
- Price above fast/slow EMA is a weak trend filter; it says nothing about volume confirmation, and the bear case shows volume momentum is negative. A trend without participation is a bull trap setup.
- No mention of the Aug 19 earnings date in the bull case — you're implicitly assuming no event risk, which is a hidden short-vol bet that can gap against you.

COUNTER TO BEAR:
- Composite z of +0.27 with rank 69/252 is not "mid-pack" — it's top 27%, and in a BULL regime catalog with zero direction-flips, that's a positive tilt, not a neutral one. Calling it low-conviction ignores that the factor model's own regime filter is green.
- The three drags (sharpe_60d, volume_momentum_60d, close_to_high_60d) are all backward-looking 60-day metrics — they measure the prior drawdown, not the current reversal. Using them to argue against a fresh bottoming signal is mixing timeframes; the technical signal is explicitly shorter-horizon.
- Insider sales at $400–$418 in June are stale — that's 10 weeks and ~8% ago. Insiders sell for diversification, taxes, or planned liquidity; without a cluster of recent sales (last 2 weeks), this is noise, not a signal.
- The Aug 10 -1.54% drop vs S&P -0.06% is one session. Cherry-picking a single down day to claim "stalling" is weak; the bull case cites two up days on Aug 7 — you're both picking single prints.
- Earnings risk cuts both ways: if guidance beats, the stock is 8% off highs with a fresh technical buy — the short setup has asymmetric upside against it. The bear case assumes disappointment without any fundamental evidence of a miss.

Technical Signal (1-XH)

action
buy
score
0.60
variant
xiaohan_gogo_v2
timestamp
2026-08-10 15:30:00
close
383.8
ema_low_fast
382.8
ema_high_slow
380.4
gogo_score
0.6
gogo_detail
2h_bottom,3h_bottom,4h_bottom,1h_blue_upper_above_yellow_upper

📈 Backtest Track Record xiaohan_gogo_v2 · signals simulated over available history (~3-month 30-min window, accumulating daily)

Trades
2
Win rate
100%
Avg / trade
+15.9%
Total (compounded)
+32%
best / worst
+30.8% / +0.9%
avg holding
34.6 days
open position
yes — marked-to-market +0.9%
Entry Exit Entry Exit Return Exit reason
2026-07-21 open 380.36 383.84 +0.9% open_position_mark_to_market
2026-03-31 2026-05-19 316.36 413.93 +30.8% adaptive_breakdown_30m

复盘 = each past xiaohan_gogo_v2 buy signal simulated: enter next bar, manage adds / adaptive sells / −25% hard stop. Demonstration only, not investment advice.

News & social (8 items)