Ticker ES
From 2010-01-18 18:00:00
To 2015-11-20 16:30:00
Security Type Future
Market USA
Resolution Minute
Status Waiting Review

It seems that in the years 2012-2015, and to a lesser extent in the years before that, there are quite some minute bars for the ES contract that have huge spikes in the High and Low in about 1 bar every day. Mostly at 16:30, which is the first bar following the break after the RTH session. 18:00 seems to be wrong often as well.

These spikes aren't present in my own tick data, meaning it's kind of throwing the wrench in my algos when I'm use high/low data.

Please see the attached algo/backtest + log for a demonstration.

Note that the heuristic used in the algo yields some false positives (where volatility was actually very high), but 90% should be correct.