Alpha Engine Research

IPO Alpha Engine

Quantitative IPO research built on the signals you can actually see before a company starts trading. Every deal scored on the same framework, with no hindsight.

Historical IPOs
Active Filings
Pre-IPO Signals
Backtest Range
Track Record

Real calls. Real outcomes.

Company
Date
Actual
What Happened
58
ANTH
Artificial Intelligence / Foundation Models
Moderate
Filed Jun 1, 2026
Oct 2026
Tracking live
$965B Series H closed May 28, making it the most valuable AI startup ever. The confidential S-1 was filed the same day as this call. Goldman, JPMorgan, and Morgan Stanley are all in talks to lead. This one is graded the moment it prices.
41
SPCX
Aerospace / AI / Defense
Cautious
Jun 12, 2026
▲ 19.3%
Called correctly
The model flagged a very rich valuation, a $4.9B net loss, and heavy reliance on a single AI compute customer, and rated the deal only Cautious. It correctly called a scarcity-driven first-day pop that would not last. SpaceX has since round-tripped from a $225 peak back to its $135 IPO price.
0
STDN
Advanced Nuclear Fuel
Avoid
Jul 16, 2026
▼ 18.0%
Called correctly
No anchor investors, a raise made up entirely of new stock, and a $79.9M accumulated deficit on just $3.4M of revenue produced the lowest score in the model's history. Standard Nuclear cut its offering by more than half the night before pricing, then opened lower and closed its first day down 18%.

Two views. One framework.

0Conviction Score
  1. 01

    Loading IPOs

    Companies with an active SEC filing or confirmed banker engagement, scored before they price. Pershing Square, Cerebras, SpaceX, Databricks, and more.

  2. 02

    Backtest

    Historical IPOs, scored using only the S-1 data that was available before they listed. See how the model held up against the actual Day 1 returns.

  3. 03

    Conviction Score

    A 0 to 100 score plus Day 1 pop and Week 1 hold probabilities, blended from binary deal signals and ratio quality, then adjusted for the current market regime.

How It Works

Pre-IPO signals, structured.

1
Binary deal signals. Eleven yes or no signals pulled from each S-1 and the deal mechanics. Things like underwriter quality, float structure, anchor investors, lockup terms, growth, and profitability.
2
Ratio quality. Gross margin, revenue growth, and cash runway scored on a continuous scale, then blended with the binary layer based on the current market regime (bull, neutral, or bear).
3
Penalty adjustments. Valuation premium, Rule of 40, debt-to-revenue, burn multiple, and net revenue retention each push the score down where the fundamentals don't justify the price.
4
Conviction and probabilities. The composite maps to one of five labels (Avoid, Cautious, Moderate, Strong, or Exceptional) along with an estimated Day 1 pop probability and a Week 1 hold probability.