Does Our Own IPO Recommendation Score Actually Work? We Checked 48 Listings
By IPO Cracker Editorial Team · 08 Sept 2026
Every open and upcoming IPO on this site carries a label from Strong Subscribe to Strong Avoid, built from a 4-factor score. We built it; we should be the first to check whether it actually means anything. So we ran it on every IPO we have tracked all the way to listing day, using only the data that existed before that IPO listed, and compared the label against what really happened.
The method
The recommendation engine combines four inputs: GMP momentum, subscription demand across QIB/NII/Retail, valuation against industry peers, and financial health from revenue growth, margins and ROE. Each factor is weighted, and missing data redistributes the weights among whatever is available.
To avoid grading the engine with information it wouldn't have had, we recomputed the score for each listed IPO using only GMP readings recorded strictly before its listing date and the final subscription snapshot from the close of bidding — the same inputs the engine would have used the day before listing. That gives a sample of 48 listed IPOs where we have both a computable score and a confirmed listing price.
The result
We grouped the 48 IPOs by the recommendation label the engine would have shown, and looked at what listing day actually paid.
| Recommendation | IPOs | Avg listing gain | % that listed positive |
|---|---|---|---|
| Strong Subscribe | 3 | +47.5% | 100% |
| Subscribe | 9 | +27.6% | 88.9% |
| Neutral | 21 | +0.6% | 42.9% |
| Avoid | 9 | -10.4% | 0% |
| Strong Avoid | 1 | -23.9% | 0% |
The ordering is exactly what it should be if the score carries real information: average listing gain falls in a straight line from Strong Subscribe down to Strong Avoid. Every Strong Subscribe IPO in this sample listed above its issue price, and every Avoid-or-worse IPO listed below it.
Where it stops being useful
The Neutral band is the honest middle: 21 of the 48 IPOs landed here, and it split close to the way a coin toss would — 42.9% positive, average gain barely above zero. A score that can separate the clear winners and clear avoids but is genuinely undecided about the middle is behaving the way a score built from real-time market signals should. GMP and subscription are noisiest exactly when the market itself is undecided.
What to do with this
- Treat the extremes as more informative than the middle. Strong Subscribe and Avoid-or-worse labels carried a clean signal in this data; Neutral did not, and shouldn't be read as a lean in either direction.
- The label summarises public data, it doesn't add new information. GMP and subscription already move the same way, as our other data posts show — the engine's job is to combine them consistently, not to know something the market doesn't.
- Still read the prospectus. A score is a starting filter across many issues, not a substitute for reading the specific risks of the one you're considering.
The honest caveats
Forty-eight is a small sample, and two of the five buckets are thin enough to be unreliable on their own: only 3 IPOs fell in Strong Subscribe and just 1 in Strong Avoid, so those two rows could easily look different with a handful more listings either way. The sample also skews heavily SME (41 of 48), so this doesn't say much yet about how the score performs specifically on Mainboard issues. And because GMP and subscription are two of the engine's four inputs, and both independently correlate with listing gains, part of what we're confirming here is that those two inputs work — not that the composite adds much beyond them.
We'll keep re-running this as more IPOs list and post the running number rather than declare victory on 47 data points.
See the live label for any current IPO on its Best IPO Today page, or check how past IPOs performed on listing.