Every Sunday night someone in a group chat posts a chart of the CME bitcoin futures with a little arrow pointing at a price above or below where spot is trading, and the caption is always some version of "gap fill incoming." It has become a reflex. The candle has a hole in it, therefore price must go back and close the hole. I used to believe it more than I do now, and the reason I stopped is that I actually started keeping track of the ones that did not fill, which turns out to be the whole game.
Why the gap exists at all
The gap is a plumbing artifact, not a signal. CME bitcoin futures trade on the exchange's calendar, which means they close Friday evening and reopen Sunday evening with a maintenance break in between, and they are shut on certain holidays. Spot bitcoin, meanwhile, never sleeps. It trades all weekend across a hundred venues. So while the futures pit is dark, spot can wander wherever it wants, and when the futures reopen, the first print can be well above or below Friday's settlement. The empty space between Friday's close and Sunday's open is the gap. It represents a price range where futures simply did not print, because nobody was there to trade them.
That framing matters because it tells you the gap is not some magnetic level the market is drawn to. It is a stretch of prices that got skipped. The "gaps always fill" idea rests on a soft intuition that price tends to revisit levels it passed through quickly, and there is something real underneath that, but it is a tendency, not a law, and the strength of the tendency depends heavily on the gap itself.
What the fill statistics actually look like
If you go back through years of these weekend gaps and tag each one as filled or unfilled, a few patterns show up that the meme completely papers over. I am going to speak in rough terms here on purpose, because anyone quoting you a precise fill percentage to the decimal is usually curve-fitting a story, and the exact number drifts depending on how you define a fill and which years you include.
- Small gaps fill more often than large ones. A tiny gap of a few tenths of a percent is almost always closed quickly, often within days, because normal intraweek chop is enough to drag price back through it. A large gap that opens after a violent weekend move is a different animal, and a meaningful share of those stay open for weeks, months, or effectively forever if the market has repriced.
- Fill rate and fill speed are two separate questions. Most gaps do eventually fill if you wait long enough, but "eventually" can mean the trade thesis is dead by the time it happens. A gap that fills after price has run 15% in the other direction first is not a win for the person who shorted into it on Monday.
- Direction interacts with trend. Gaps that sit against the prevailing trend tend to get filled faster, because the trend is already pushing price back toward them. Gaps that open in the direction of a strong trend can hang unfilled for a long time, because the market keeps walking away from them.
The honest summary is that fill rate is high if your holding horizon is infinite and your invalidation is nonexistent, and it drops fast once you attach a real stop and a real timeframe. That gap between the headline stat and the tradeable stat is where most people lose money.
The survivorship trap
Here is the part that took me too long to internalize. The gaps you remember are the ones that filled in two days and looked clean on the chart. Those get screenshotted and posted. The gap that opened after a big weekend rally and never came back does not get a victory-lap post, so it quietly falls out of everyone's memory. Your sample is being filtered by what was satisfying to share, and a sample filtered that way will always make the pattern look stronger than it is.
The fix is boring and it works. Log every gap, filled or not, before you know the outcome. Write down the gap size, the direction, the date it opened, and then whether and when it filled. After a few months of that you will have a base rate that belongs to you instead of to the group chat, and it will almost certainly be less exciting than the meme promised. Less exciting is exactly what you want, because it is the truth.
A rules-based way to actually trade it
If you still want to trade toward an unfilled gap, the trade has to have a defined invalidation or it is just hope with a chart attached. Here is the frame I use, and you can adjust the thresholds to your own logged data.
- Only bother with mid-size gaps. Skip the tiny ones because they fill on their own and there is no edge to capture, and be very skeptical of the enormous ones because those are the ones that stay open. The interesting range is the middle.
- Trade with the pull, not against it. Prefer a gap that sits against the recent trend, so the market's own drift is helping you toward the fill rather than fighting you.
- Define the entry as a level, not a feeling. Enter when price starts moving toward the gap from a structure you can point to, not just because it is Monday and the gap is there.
- Set invalidation on the far side of the gap's origin. If price closes decisively beyond the level that created the gap, the thesis is wrong. That is your stop, and it needs to be sized so a single miss does not hurt.
- Put a clock on it. Give the trade a fixed number of days to work. If the gap has not filled by then, close it. An unfilled gap that has sat for weeks is telling you the market repriced, and you should listen.
The clock rule is the one people skip, and it is the one that saves you. Without it, a gap trade quietly turns into a bag you hold forever while telling yourself it "still has to fill." It does not have to do anything.
Where this fits
Gap fills are a small, situational edge at best, and they are far more useful as context than as a standalone system. A visible unfilled gap below is one more reason to be cautious about chasing a rally, and a gap above is one reason a dip might have somewhere to go. I treat it as one input among several, alongside where the whales are positioned and what the futures basis is doing, which is roughly how the scorecards on Blockcircle are built, by stacking a few honest signals rather than betting everything on one tidy pattern.
Keep the log, respect the clock, and remember that the clean fills are loud and the stuck gaps are quiet. Once you can see both halves of the sample, the trade gets a lot less magical and a lot more useful.