Validation
A quote is a promise. The check is to hold the size resting at the touch against the volume that actually printed in the five minutes after it. Across 273 bars the median was 21x - far more traded than was ever displayed, so the book refills and the estimator is conservative.
Most of this universe could not be validated, and that is the result. Before comparing anything, the check sums twenty-four hourly bars for a symbol and holds the total against that symbol's own rolling 24h turnover. On crypto the two land within 2% of each other. On tokenized stocks they diverge by a factor that differs per symbol - base volume and quote volume diverge by the same factor, so it is not a units error, and it is not constant, so it is not a fixed multiplier. Which feed is right cannot be settled from outside, so 6 stock symbols were excluded rather than averaged over.
| Excluded symbol | Candle vs ticker | Why |
|---|---|---|
| RNVDAUSDT | 40.2x | candle volume is 40.2x the ticker's 24h turnover |
| RTSLAUSDT | 38.2x | candle volume is 38.2x the ticker's 24h turnover |
| RAAPLUSDT | 51.3x | candle volume is 51.3x the ticker's 24h turnover |
| RMSFTUSDT | 37.4x | candle volume is 37.4x the ticker's 24h turnover |
| RSYKUSDT | 44.9x | candle volume is 44.9x the ticker's 24h turnover |
| RPBRUSDT | 101.0x | candle volume is 101.0x the ticker's 24h turnover |
A ratio above 1 has one explanation; below 1 has two. More printing than was displayed can only mean the book refilled. Displayed size going unconsumed could mean it was withdrawn before anyone hit it, or that nobody wanted it. Idle and phantom liquidity look identical from outside, and this data cannot separate them.
A symbol is only compared once its own two volume feeds agree with each other. Anything outside a 0.80 to 1.25 band is excluded and named above rather than folded into an average. Publishing a ratio computed from a feed already shown to be unreliable would be worse than publishing nothing.