StakeStats
10SC Free

Seed and Nonce Errors: Why Your Verification Won't Match

Verification not matching? Learn the most common seed and nonce errors, from hashed seeds to off-by-one counts, and how to fix each one step by step.

By Seal · 2026-11-13

Seed and Nonce Errors: Why Your Verification Won't Match

Seed and Nonce Errors: Why Your Verification Won't Match

Player checking a completed bet on laptop

Most seed and nonce errors are input mistakes, not rigged games. The usual causes are the hashed server seed pasted where the revealed one belongs, a nonce that is off by one, a stray space in a copied value, or the wrong game selected. Check those four in that order and most mismatches resolve in a few minutes. Our server seed rotation guide walks through getting a clean, revealed seed to test with, and our verification tools on Stake Stats take server seed and nonce inputs for bet verification.

TL;DR

  • A verified result is built from the unhashed server seed, your client seed, the nonce of that specific bet, and the right game. Change one character and the output changes completely.
  • The most common errors: using the hashed seed, using the current nonce instead of the bet's nonce, copy-paste whitespace, and nonce counting that starts at 0 versus 1.
  • A nonce can legitimately repeat after a seed rotation, because it only counts bets within one seed pair.
  • If the random event matches but the payout does not, the problem is in game settings or payout rules, not the seeds.
  • Save the hash, seeds, nonce and screenshots before you do anything else. Rotating again can make the evidence harder to rebuild.

Table of Contents

What the three values do

You need the roles of the inputs straight before you can debug them.

The server seed is the casino's secret random value. The client seed is your contribution, either a string you choose or one generated for you. The nonce is a counter that goes up with each bet. According to the Provably Fair Foundation, single-player provably fair games usually generate results from a SHA-256 hash of the server seed, client seed and nonce combination. The plain server seed must be shown to the player once the game has finished or the seed pair has been rotated.

That last point explains the first error most players hit. Before you rotate, you only see a hash of the server seed. The hash is a commitment: it proves the casino fixed its seed before you bet, but you cannot replay a bet from it.

On Stake specifically, each verifiable bet uses four inputs: server seed, client seed, nonce and a cursor. According to a summary of Stake's provably fair page, the server seed is a random 64-character hex string, the nonce rises by 1 with each new bet, and the calculation uses HMAC_SHA256 to produce 32 bytes of output. The cursor matters in games that need more random numbers than one block of output provides, such as multi-step or multi-pick games.

Server seed and bet inputs flow into HMAC output

The seed and nonce errors that cause most mismatches

Here is the list in the order we would check it. The earlier items are the likeliest.

1. Using the hashed server seed instead of the revealed one

The hashed seed and the revealed seed are different strings. The hash is what you see while the seed pair is active. The revealed seed appears only after rotation. If you paste the hash into a verifier as if it were the seed, the output will not match, and nothing will tell you why.

How to tell them apart: both can look like long hex strings, so do not go by appearance. Go by where you got them. If the seed pair is still active, you have the hash. If you rotated, open the previous seed pair in your fairness or seed history and copy the unhashed value from there. Then confirm it: hash the revealed seed with SHA-256 and check that it equals the hash you saw before you started betting. If it does not, you have the wrong seed pair, and no amount of tinkering with the nonce will fix it.

2. Using the current nonce instead of the bet's nonce

The counter keeps rising as you play. The nonce you need is the one attached to the completed bet, not the number shown now. A bet you placed 40 rounds ago used a nonce 40 lower than your current one.

This is the classic off-by-one family. If you copy the nonce from the live counter after another bet, you are off by at least one. Always take the nonce from the bet's own record in your bet history.

3. Zero-based versus one-based counting

Some systems start the nonce at 0, others at 1. The first bet on a new seed pair is nonce 0 on one platform and nonce 1 on another. If your verifier assumes the wrong convention, every result will be wrong by exactly one position, which is why this one is so confusing: the numbers look almost right.

Test it with the first bet after a rotation. If the bet history shows its nonce, enter exactly that. If your verifier adds or subtracts one on its own, switch that setting off or pick the platform-specific mode.

4. Wrong client seed

If you changed your client seed during a session, the bets before and after the change belong to different seed pairs. A bet placed under client seed A will not verify under client seed B, even with the right server seed and nonce. Check the client seed recorded in the bet's own details, not the one currently active in your settings.

This is also the reason to set your own client seed deliberately and write it down. If you leave the default, it is easy to lose track of which value was live at the time of a particular bet.

5. Whitespace and copy errors

A trailing space, a line break or a missing last character will break a hash completely. Hash functions are built so that a one-character change produces an unrelated output. Paste each value into a plain-text field first, trim the ends, and compare lengths. A Stake-style server seed should be 64 hex characters, so a count of 63 or 65 means a copy error.

6. Text versus hexadecimal bytes

Some verifiers treat the seed as plain text, while the game may treat it as hex bytes, or the other way around. The same string yields two different HMAC keys. If everything else checks out, this is the next thing to test. Follow the exact encoding the game's official verification instructions describe.

7. Wrong separators, wrong field order, wrong key

In an HMAC, one value is the key and another is the message. Swapping them gives a valid-looking but wrong result. The same goes for the separator between client seed, nonce and cursor, and for the order of those fields. Stake-style schemes put the server seed in the key position and build the message from the client seed, nonce and cursor joined by colons. A different platform may do it differently, so match the platform you played on.

8. Wrong game, mode or version

The same random numbers become different outcomes in different games. A dice roll, a mines grid and a slot spin each convert the raw numbers in their own way. Even within one game, the risk level, number of segments, number of mines or a version of the game can change the mapping. Choose the exact game and settings you played. A mismatch here looks like a seed error but is a setup error.

9. Rounding and display precision

The game may show a result rounded to two decimals while the raw calculation produces many more digits. If your verifier gives 1.9999 and the game shows 2.00, that is a display difference. Keep the raw value so you can see which it is.

Quick reference: symptom to likely cause

What you see Likely cause What to check
Hash of your revealed seed does not equal the pre-bet hash Wrong seed pair or hashed seed pasted Pull the seed from the correct rotated pair
Result is wrong but looks "close" in sequence Nonce off by one 0 versus 1 start, bet's own nonce
Everything is wrong, nothing resembles the game Wrong client seed or encoding Client seed on that bet, text versus hex
Random event matches, payout differs Settlement, not randomness Game settings, paytable, rounding
Only a few old bets fail Seed changed mid-session Which seed pair each bet belongs to

Why the same nonce shows up again

Players sometimes report that they saw nonce 0 or 1 again and assume something broke. It is normal. The nonce is scoped to a seed pair. Galabet's documentation explains that after rotation the nonce can restart because the new HMAC key makes the calculation unrelated to the old pair.

So a nonce is not a unique bet ID. It is a fairness counter within one pair of seeds. When you look up a bet, you need the full set: the revealed server seed, the client seed and the nonce, together with the game. Any one of them alone does not identify the bet.

This also covers a common gambler's-fallacy worry. According to the summary of Stake's provably fair page, each bet is independent, and results do not change with previous wins, losses, time played or wager size. The nonce only keeps bets distinct. A long losing run does not make a win "due", and the counter does not track one.

A step-by-step way to reproduce a bet

Work through this once, in order. Do not skip the commitment step.

  1. Save the commitment before you bet. Copy the hashed server seed shown for your active seed pair and keep it somewhere you control.
  2. Record the bet. Note the bet's client seed, nonce, game, settings and visible result. A screenshot of the bet details is good evidence.
  3. Rotate to reveal the seed. Rotating the seed pair is what reveals the unhashed server seed. Our guide to rotating a server seed and verifying your bets covers when to do it and what to save first.
  4. Verify the commitment. Hash the revealed server seed with SHA-256 and confirm it equals the hash you saved. If it does not, stop here.
  5. Clean the inputs. Trim whitespace, check the 64-character length, and confirm the client seed and nonce are those of the bet.
  6. Choose the exact game and settings. Match the mode, risk level and version.
  7. Run the calculation and compare the event. Compare the raw outcome (the roll, the grid, the multiplier) before looking at the payout.
  8. If it still fails, change one variable at a time. Try nonce plus or minus one, then text versus hex, then separators. Changing several things at once hides which fix worked.

Our tools on Stake Stats are built around this flow: you enter the server seed and nonce, and the bet is verified without you writing the HMAC code by hand. That takes the encoding, separator and field-order mistakes in steps 5 to 7 out of your hands.

Event matches, payout does not

A separate class of problem is when the recomputed event matches the game but the amount paid looks different. That is not a seed or nonce error at all. The randomness did its job, and the question is how the game turned the event into a payout.

Things to check:

  • The paytable or multiplier table for that game and version.
  • Your stake, any bet conditions and any bonus rules that applied.
  • Maximum win limits and rounding.
  • Action history for games with several steps, where your choices affect the payout.

For slots, the multiplier tables and the long-run statistics are worth reading together. Our guides to the Stake hit rate list and the Stake volatility list show how to read hit rate and volatility so you know what a normal spread of results looks like before you decide a single result is odd.

What a successful check proves

A match proves several useful things. The revealed seed matches the commitment you saw before betting, so the casino could not have changed it after seeing your bets. Combined with your client seed, the result was fixed by values neither side could fully control alone. And you can reproduce the outcome yourself rather than trusting a badge on screen.

It is best to keep a proper record: the pre-round commitment, the disclosed inputs, the game and version, your calculated output and the displayed result. A single screenshot of a "fair" label is thin evidence by comparison.

A check on one round confirms that round. Checking a sample of rounds across a seed pair builds a fuller picture, and a verification tool that handles many bets saves the manual work.

If it still does not match

If it still does not match

First, a mismatch is not proof of manipulation. It means something in your reproduction differs from the recorded system. Recheck input order, which value is the key and which the message, text versus hex bytes, nonce and cursor values, formula version and rounding.

If the exact official process still fails, preserve everything: screenshots, the hashed seed, the revealed seed, client seed, nonce, game and timestamps. Then contact the platform's support with those raw values. Do not rotate again or edit your seeds before you have saved them, and do not trim the evidence down to a summary. Support can only investigate what you can show.

How we approach seed and nonce checking

Most of the verification questions we get come down to one thing: people treat the check as a single yes or no, when it is really a chain of small inputs that all have to be right. That is why we built our tools around server seed and nonce verification, so the formula, encoding and ordering are handled the same way every time and you can focus on whether you copied the right bet.

Our view is that a provably fair claim is only as useful as your ability to test it. That means saving the commitment first, using the bet's own nonce, and comparing the raw event before the payout. We pair verification with real-time RTP, volatility and hit rate stats across 2500+ Stake Engine games, so you can tell the difference between a verification problem and an ordinary run of variance.

The habit we would like every player to build is simple: record, rotate, verify, compare. It takes a few minutes and tells you far more than a badge does.

Ian

Check your bets with Stake Stats

If you would rather not rebuild the calculation by hand, Stake Stats offers server seed and nonce bet verification, plus max win and bet replay lookup, alongside live RTP, volatility and hit rate data for 2500+ Stake Engine games. Enter the revealed seed, client seed and nonce, pick the game, and compare the result with what you saw.

Stakestats

Start at stakestats.net, and read the seed rotation guide first if you have not yet revealed a seed.

FAQ

Why does my verified result not match the game?

The usual causes are the hashed server seed used in place of the revealed one, the wrong nonce, a copy error, or the wrong game settings. Check them in that order. A mismatch alone does not mean the game was unfair.

Can the same nonce appear twice?

Yes, across different seed pairs. The nonce counts bets within one pair, and after you rotate, the count restarts with a new server seed. Within a single seed pair, each nonce is used once.

Does the nonce start at 0 or 1?

It depends on the platform. Use the nonce shown in the bet's own record, and check whether your verifier assumes a zero-based or one-based start. If every result is off by exactly one position, this is the likely cause.

Why can't I verify my current bets?

The active server seed stays secret until the seed pair is rotated. Until then you only have its hash. Rotate the seed pair, take the revealed server seed, confirm its SHA-256 hash matches the one you saved, then verify your bets.

What should I do if I think a result was wrong?

Save the hashed seed, revealed seed, client seed, nonce, game, settings and screenshots. Rerun the check with the exact official process, changing one variable at a time. If it still fails, send the raw values to the platform's support.

Sources