/volume/history returning inconsistent results

Over the past few days I’ve noticed requests to /volume/history returning different results.

For example, the volume (spot + derivative) for June 14 was previously returning 400b+ (which seems correct as it was a high volume day), but now comes back as 122,732,917,503.

I paid for the nomics API only for this data and this is the third time that this endpoint has been messed up. I’d love to keep using nomics, but unless these numbers start having consistency, I will have to go elsewhere :frowning:

Also, the volume (spot + derivative) number returned for 2022-01-27 is 230,469,965,865,845 which is definitely wrong

Thanks for the report, we found and fixed an outlier market causing the spike on 1/27.

Did you look into the current result for 2022-06-14?

June 14th is 191b which is correct.

The data returned continues to change depending on when I query.

About 8 hours ago, for the range 2022-06-09 - 2022-06-15 I got:

|2022-06-15|266001501493|
|2022-06-14|192787406517|
|2022-06-13|286215699511|
|2022-06-12|250694464564|
|2022-06-11|212977445940|
|2022-06-10|218494262797|
|2022-06-09|170712306347|

Just now, I made the same request and got:

|2022-06-15|270821319207|
|2022-06-14|334014997536|
|2022-06-13|443755026380|
|2022-06-12|255342829361|
|2022-06-11|212977445940|
|2022-06-10|218494262797|
|2022-06-09|170712306347|

The aggregate volume for days 6-13 - 6-15 is considerably different in the second request. Why would the volume number change depending on when the request was made?

Btw I think the second results are the accurate numbers if that helps with debuging. 6-13, 6-14, 6-15 were very high volume days.

The spike lines up with what coingecko binance spot volumes show:

We are constantly scanning and adjusting for market outliers as well as constantly adding or mapping new markets. So for example we could have a market with 7 days of data and today we determined which asset it was. At that point, we count it towards global stats historically. Another example is when we determine a market is mismapped or reporting outlier data, so we stop including it in global statistics.

We can be more accurate as more time passes since we can scan more data to determine what is an anomaly vs normal data patterns.

Ah ok, makes sense, appreciate the explanation. I can sympathize with the complexity of getting an accurate picture of total market volume.