Field Reference

What the numbers mean. Parameters and schemas are in the CMI API Reference.

Price and activity

open high low close — first, highest, lowest and last trade price in the interval, per venue. Because high and low are extremes, the range widens simply as more trades happen, so trade count is worth checking before a wide range is read as volatility.

vwap — average price weighted by the amount traded at each price. It shows where the weight of trading actually sat, which the close does not.

volume — total traded quantity in the base currency. Multiply by vwap for a quote-currency figure.

trades — number of executions. Nearly every other statistic degrades as this falls, so it works as the denominator of confidence in everything else.

volat — dispersion of trade prices within the interval, in price units rather than percent.

Order flow

buysell — share of trades on the buy side, bounded 0 to 1, so neutral is 0.50. Values around 0.4–0.6 indicate ordinary balance. Its variability carries more information than its level: organic markets wander, while one pinned near 0.50 minute after minute is being held there.

buysellvol — the same share weighted by quote value instead of trade count. Most useful alongside buysell: when the two agree, buys and sells arrive in similar sizes, and a gap between them means they do not.

avgsize — average trade size in base currency.

ttdist — 60 counts, one per second of the minute, showing when trades landed. Bursty is normal and healthy; what stands out is a distribution too even to be chance, which suggests mechanical pacing. Measure the variability rather than the flatness.

Distributions

vdist — a histogram of trade notionals (price × amount) as [bin index, count] pairs: 100 equal-width buckets spanning that minute's own smallest to largest trade. Because the bucket count and span are fixed relative to the window, the shape of the count vector is scale-free and comparable across venues and time, which is enough for a uniformity measure.

Heavy concentration in bin 0 is the healthy case. An organic market's largest trade sits far above its typical trade, so the body compresses into the bottom of the range; around 95% in bin 0 is normal. Counts spreading across many buckets mean the maximum sits close to the typical size, which is what repeated sizes look like.

fdist — counts of leading digits 1–9, taken from each trade's quote notional. Notionals span several orders of magnitude, which is what digit analysis requires.

btest — the largest gap between an observed and an expected digit share. Lower means closer conformity, so a low value is unremarkable. It is computed per minute on however many trades that minute held, which is usually too few to detect much; pooled across hours it works well.

vvcorrel — correlation between volume and volat across the hour's minutes. With 60 observations it is noisy; the level of volume per unit of price movement, compared across venues, is more robust.

Cross-market

disc — which venue printed the interval's highest and lowest price, and how far each sat from the cross-market average, in percent. Both values arrive positive, measured in opposite directions, and at most two venues are reported per interval. Gate on the volume behind the print before ranking by size.

Availability

Field group	1m	1h / 1d	Cross-market
ohlcv, vwap, trades	✅	✅	✅
avgsize, volat, buysell, buysellvol, btest	✅	✅	❌
fdist, vdist, ttdist	✅	❌	❌
vvcorrel	❌	✅	❌
disc	—	—	✅

A field unavailable at the requested granularity is simply absent from the response, so clients should check field presence rather than rely on the status code.