do you drop stalled bars or just avoid those symbols?

Read the research post on frozen crypto quotes over a full day. The part that got me was that a stall can last the whole recording, not just a few minutes. My question is practical. In a backtest, what do you actually do with those minutes? Options I can see: 1) drop the bars entirely and let the strategy sit out 2. keep them but mark the symbol untradeable for that run plus some buffer after 3) filter the universe up front so thin stuff never enters Option 3 feels like cheating on the research, since the post says a majors-only list would barely show the problem. But option 1 makes a mean reversion test look nicer than reality, because the fake flat stretch never triggers an entry and never eats a stop. Also unclear to me: after a stall ends, is that first tick a real price or a catch-up jump? If it's catch-up, half my breakout signals in the low-liquidity names could just be data waking up. What do people here do?
L
LurkerLee
9 posts · 0 followers
+ Follow

Related reading

2 replies

DataDrivenDee· Aug 2026 ago
Fair questions. On the catch-up thing — how would we test that? My instinct is to take every stall end, look at the return on the first live bar versus a matched sample of normal bars on the same symbol, and see if the distribution is fatter. If it is, that's your answer without needing tick data from another venue.
I
IndicatorSkeptic· Aug 2026 ago
Option 3 isn't cheating, it's just admitting you were never going to trade those names anyway. Half this hobby is inventing sophisticated ways to backtest instruments you'd never actually touch with real money, then acting surprised when the fill doesn't exist.
Sign in to reply →
← All strategies & backtesting