Track Record: Week of April 26, 2026

106 signals tracked. One signal type is crushing it. Another is quietly bleeding alpha — and we're keeping both.

After tracking 106 signals against SPY across six signal types, the data is screaming something most retail investors would find backwards: Reddit sentiment calls are outperforming SEC insider filings by a country mile. Our Reddit long calls are delivering +2.40% mean alpha (excess return vs SPY) at the 10-day mark with a 50% win rate. Meanwhile, Edgar insider buys tagged as "HIGH confidence" — the kind Wall Street venerates — are sitting at a 17% win rate and -5.56% mean alpha at 30 days. If you'd blindly followed every "smart money" insider buy signal we flagged this quarter, you'd be underwater vs simply holding SPY.

The Scoreboard

Here's where we stand across all signal types, measured at each type's native evaluation window:

The headline finding: Reddit sentiment — when filtered through our threshold model — is generating measurable alpha. Insider filing signals are not. At least not yet, and not at these sample sizes.

The Winner: Reddit Long Calls

Reddit long calls are the clear performance leader, and it's not close. At +2.40% mean alpha over 26 signals, this signal type is doing exactly what we designed it to do: capture explosive short-term moves driven by coordinated retail attention and momentum ignition.

The conceptual edge here is velocity. When a ticker crosses our sentiment threshold on Reddit (measured via comment volume spikes, upvote velocity, and keyword density in high-traffic subs), it often means a catalyst has already begun propagating through retail networks. We're not trying to front-run institutional money — we're trying to ride the second wave of a retail-driven liquidity event before it peaks.

Real examples: $AMC signaled on March 31st and delivered +57.6% alpha by day 26. This wasn't a meme-stock resurrection — it was a short-term earnings speculation play amplified by WallStreetBets coordination around options expiry. $INTC signaled April 8th, returned +34.4% alpha at T+18 on the back of unexpected foundry contract news that Reddit caught before the mainstream financial press did. $SMCI on March 21st gave us +31.6% alpha at T+36 as AI infrastructure hype cycled back through retail channels.

The pattern is consistent: these aren't buy-and-hold value plays. They're short-duration momentum trades where crowd psychology creates temporary mispricings. Our 10-day evaluation window is deliberate — we're capturing the spike, not the reversal.

Win rate is exactly 50%, which tells us the model isn't just getting lucky on a few outliers. Half the time we're right, half the time we're not, but the winners are big enough to overcome the losers. That's an edge.

The Loser: Edgar Insider Buys (HIGH Confidence)

Now for the uncomfortable truth: our highest-confidence insider buy signals are getting destroyed. Twelve signals tracked, 17% win rate, -5.56% mean alpha. If you'd followed every one of these "smart money is buying" calls, you'd have underperformed SPY by an average of 5.56% per trade at the 30-day mark.

Why are we still tracking them? Two reasons.

First, sample size. Twelve signals is not enough to declare the hypothesis dead. Insider buying is one of the most academically validated signals in finance — dozens of peer-reviewed studies show that clustered insider purchases, especially by C-suite executives, correlate with forward returns over 6-12 month windows. Our problem might be the window (30 days may be too short) or the threshold (our "HIGH confidence" filter may be miscalibrated). We won't know until we have 50+ signals.

Second, intellectual honesty. If we only reported winners, we'd be another hype newsletter. The entire purpose of this track record page is to show you what's actually happening, not what makes us look smart. Insider buys are struggling. We're disclosing that openly because credibility matters more than marketing.

The worst individual call in this category: $LULU on March 24th, down -20.0% alpha at T+33. The CEO bought $2M in stock two days before an unexpected inventory write-down was leaked. Either the buy was a scheduled 10b5-1 trade (which we should filter better) or the executive genuinely misjudged timing. Either way, our model took the signal and lost.

We're continuing to track these, but we're also experimenting with longer evaluation windows (T+60, T+90) to see if the alpha shows up later. Insider buying might just be a slower burn than Reddit momentum.

Methodology Note

Every number in this post comes from our live tracking database. Alpha is calculated as the stock's return minus SPY's return over the same calendar window, adjusted for signal direction (long calls measure outperformance, short calls measure underperformance). Full methodology, raw data exports, and signal history are available at crawdrift.com/track-record. We do not backtest. We do not cherry-pick. If we flag a signal publicly, it goes into the tracker, and we report it here — win or lose.

What We're Learning

This week's data is forcing us to confront an uncomfortable hypothesis: maybe retail sentiment is a better short-term signal than insider filings in 2026's market structure.

Why might that be true? Two theories:

1. Regulatory lag. Insider transactions must be filed with the SEC within two business days, but the decision to buy often happens weeks earlier. By the time we see the Form 4, the information edge has decayed. Reddit sentiment, by contrast, is real-time. When a subreddit explodes around a ticker, we see it as it happens.

2. Liquidity regime. We're in a market where retail options flow can move stocks intraday. A coordinated Reddit push on a mid-cap name can trigger gamma squeezes, market-maker hedging, and momentum algos — all of which create short-term alpha opportunities that have nothing to do with fundamentals. Insider buying signals fundamentals. Fundamentals might not matter on a 10-day window anymore.

We don't know yet if these theories hold. But we're adjusting the model in two ways:

• Expanding Edgar windows: We're now tracking insider signals at T+60 and T+90 in addition to T+30. If the alpha is there, it might just be slower.
• Tightening Reddit thresholds: With a 50% win rate and strong mean alpha, we're testing whether raising the sentiment threshold (requiring more velocity to trigger a signal) improves the win rate without killing the alpha.

Small sample caveats apply everywhere. We have one Edgar cluster buy signal on record. One. That's not a trend, it's a data point. We'll keep tracking.

Get Next Week's Signals

This post shows you what happened. Our weekly newsletter tells you what's happening now — which tickers just crossed our thresholds, which signal types are firing, and how we're playing them.

No hype. No guarantees. Just the same data-driven approach you see here, delivered every week.

Subscribe at crawdrift.com.