ψ Psi Quant Research

Signal Report · US

CLSK HOLD

Composite decision report grounded in 1-XH strategy signals and a completed multi-agent debate (agents-3node-v1), generated at 2026-09-15 10:03.

Final Agent Decision

Action
HOLD
Weight
0%
Confidence
45%
Variant
xiaohan_gogo_v2

Reasons

Bull Case (Analyst)

- Technical signal is a fresh `buy` (score 0.60, xiaohan_gogo_v2) with a multi-timeframe bottoming pattern: `2h_bottom, 3h_bottom, 4h_bottom` plus `1h_blue_upper_above_yellow_upper` — a momentum-turn setup, not a chase.
- Cross-sectional composite z = +0.45, rank 39/252 (top 15.5%), and the catalog is explicitly **BULL** with 0 regime direction-flips — the factor regime is aligned with the long side.
- The dominant lifter is `xh_gogo_score` (+2.19), which is the same family as the technical signal — the cross-section independently corroborates the timing signal rather than contradicting it. Secondary lifters `beta_60d` (+0.93) and `mean_revert_20d` (+0.90) both favor a high-beta, recently-pulled-back name in a risk-on tape.
- News flow has concrete catalysts: August update of 593 BTC produced, a "massive lease deal" cited as driving trader interest, and a +6.83% session on Sept 11 with data-center optimism supporting shares.

Bear Case (Analyst)

- Not in the production basket — rank 39/252 is good but not top-decile, so the cross-section did not confirm this as a high-conviction long. The technical `buy` is not basket-corroborated.
- Draggers are structural, not noise: `drawdown_3m` (-0.97), `downside_vol_60d` (-0.95), `sharpe_60d` (-0.60). The name has been a poor risk-adjusted performer with heavy left-tail volatility — exactly the profile that punishes a 0.60-score timing entry.
- The bull lifters are internally contradictory: `mean_revert_20d` (+0.90) and `beta_60d` (+0.93) are short-horizon/risk-on factors, while the drag set is medium-term quality/risk. This is a bounce trade inside a damaged trend, not a trend reversal.
- News is mixed-to-negative on the fundamentals: "losses deepen," insider selling of US$1.4m over the past year, and a session where CLSK sank 6% while BTC rose 7% — clear relative weakness versus its own underlying driver.
- Price 12.61 sits below both `ema_low_fast` and `ema_high_slow` at 12.91 (fast = slow, i.e. no bullish EMA separation), so the "bottom" call is unconfirmed by the moving-average structure.

Caveat: the bundle has no order-flow, funding, or options data, and the news items are headline-level with no timestamps beyond dates — so neither side can be sized off positioning evidence.

Devil's Advocate

COUNTER TO BULL:
- `xh_gogo_score` (+2.19) is the dominant lifter *and* the same family as the technical signal. That's not independent corroboration — it's the same factor counted twice. The "cross-section confirms timing" claim collapses if the composite is largely a re-weighting of the signal you're already long.
- Rank 39/252 (top 15.5%) is being sold as alignment, but the catalog being BULL with "0 regime flips" is a *catalog-level* statement, not a name-level one. A bull regime lifts everything; it says nothing about whether *this* name is the right expression.
- "Multi-timeframe bottoming" (2h/3h/4h bottom) is a pattern label, not evidence of a turn. Bottoming patterns fire repeatedly in downtrends — that's how you get the `drawdown_3m` and `downside_vol_60d` drags in the first place. No base confirmation, no volume, no higher-high.
- `1h_blue_upper_above_yellow_upper` is a single-TF momentum cross. Calling it "momentum-turn, not a chase" is assertion; a 0.60 score is mid-conviction, and mid-conviction timing entries into damaged trends are the classic false-positive zone.
- News catalysts are stale and unquantified: "August update," "massive lease deal" (no size, no counterparty, no terms), and a Sept 11 session. No timestamps, no follow-through. A +6.83% day is as easily a squeeze as a re-rate.
- `beta_60d` (+0.93) and `mean_revert_20d` (+0.90) as "risk-on tape" lifters assume the tape stays risk-on. That's a regime bet smuggled in as a stock thesis.

COUNTER TO BEAR:
- "Not in the production basket" is a threshold artifact, not a verdict. Top 15.5% with a fresh signal and aligned regime is a legitimate long candidate; basket membership is a construction choice, not a truth.
- The drag set (`drawdown_3m`, `downside_vol_60d`, `sharpe_60d`) is *backward-looking*. Mean-reversion entries are supposed to look bad on trailing risk metrics — that's the entry condition, not a disqualifier. Using them to veto is circular.
- Calling the lifters "internally contradictory" overstates it. Short-horizon reversion + high beta is a coherent bounce profile; the bear is asserting the trend must dominate without showing trend evidence beyond EMAs.
- The EMA point cuts both ways: fast = slow at 12.91 with price 12.61 is *compression*, which precedes expansion. "No separation" is not "no turn" — it's undecided, and the bear is reading indecision as bearish.
- News is cherry-picked: "losses deepen" and insider selling are dated; the CLSK-vs-BTC divergence is one session. One day of relative weakness is noise, and insider selling over a year is a weak signal at best.
- The bear's own caveat (no order-flow, funding, options) applies symmetrically — so the confident "bounce inside damaged trend" framing is also unsupported by positioning data.

Technical Signal (1-XH)

action
buy
score
0.60
variant
xiaohan_gogo_v2
timestamp
2026-09-15T15:30:00-04:00
close
12.61
ema_low_fast
12.91
ema_high_slow
12.91
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
50%
Avg / trade
-2.3%
Total (compounded)
-5%
best / worst
+5.8% / -10.4%
avg holding
19.4 days
open position
yes — marked-to-market +5.8%
Entry Exit Entry Exit Return Exit reason
2026-08-10 open 11.92 12.61 +5.8% open_position_mark_to_market
2026-08-03 2026-08-06 14.61 13.09 -10.4% adaptive_breakdown_1h

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)