Every cycle the same screenshot makes the rounds. The 50-day moving average curling up through the 200-day on the bitcoin chart, a big arrow, a caption about how this exact signal called the last bull market. And every cycle a wave of people buys the cross, sits through an ugly pullback, and quietly concludes that technical analysis is astrology with extra steps. The excitement and the disappointment come from the same misunderstanding about what a crossover actually measures.
The mechanics first, for anyone newer to this. A golden cross is the 50-day moving average closing above the 200-day. A death cross is the reverse, and that is the entire indicator. It contains no information that was not already sitting in the last two hundred daily closes, which makes it a summary of the recent past by construction. People trade it like a forecast even though it behaves more like a weather report that shows up six weeks late.
What the record actually shows
I have backtested this crossover on bitcoin, ethereum, and the major US indices enough times that I trust the shape of the results, even though the exact numbers shift depending on your start date and whether you demand a daily close or count intraday touches.
On bitcoin, the golden cross has historically caught every major bull market. That sounds great until you look at where the entries landed. Bitcoin turns violently off its lows, so by the time the 50-day climbs through the 200-day, price has typically already recovered half or more of the ground from the bear market bottom. The lag between the actual low and the signal has historically run weeks to months. And after the cross prints, a pullback in the following weeks is common, because the signal tends to fire into short-term overheated conditions. Buying the golden cross and immediately sitting through a double-digit drawdown is close to a rite of passage.
Ethereum shows the same pattern with more noise. Higher volatility means the two averages tangle more often, so you get more whipsaws, worse average entries, and a couple of signals per cycle that simply fizzle. The crossover still separates bull regimes from bear regimes reasonably well, it just charges more friction for the service.
The indices are where the popular narrative really falls apart. On the S&P 500, the death cross has historically been close to a coin flip as a sell signal, and a surprising number of them printed near local bottoms rather than ahead of further damage. The logic is simple once you see it. The 50-day only falls through the 200-day after weeks of sustained selling, and weeks of sustained selling in a market that structurally drifts upward often means the panic is nearly exhausted. Golden crosses on indices look better in the stats, but a lot of that is the long upward drift of equities doing the work rather than any timing skill in the signal itself.
The lag comes from the arithmetic
What frustrates people is that the lag cannot be tuned away. Every moving average system trades lag against noise. Shorten the windows and the cross arrives earlier but fires constantly in chop. Lengthen them and the noise disappears but confirmation shows up months after the turn. The 50/200 pair is one point on that curve, a deliberately conservative one, chosen because it changes its mind rarely.
A 200-day average is still carrying prices from more than six months ago, so for the 50 to drag itself across it, the new trend has to persist long enough to overwhelm all that history. On an index that turns slowly this costs you a modest slice of the move. On bitcoin, where a bear can flip to a bull inside a quarter, confirmation routinely arrives after a large part of the repricing has already happened. Anything you use for entry timing needs to be early or at least on time, and a crossover is structurally late every single time it fires. That lateness is tolerable if your holding period is months or years. It is fatal for swing entries, which is why the tight-stop crowd gets chewed up by it. The signal operates on a timescale of months, the stop operates on a timescale of days, and mixing the two guarantees you get shaken out of positions the signal was actually right about.
Use it to size, not to enter
Where the crossover has earned its keep, both in my own testing and in the long-horizon trend research I take seriously, is as a regime filter. The question it answers well is what kind of market you are in, one that rewards carrying risk or one that punishes it. In the historic crypto bears, the first leg down happened before the death cross, but the deep grinding majority of the damage came after it printed, while the 50 sat below the 200 for months. A filter that keeps you small through those stretches does not need good entries to be worth having. Studies of trend filters across asset classes tend to land in the same place, roughly similar returns to buy and hold, with meaningfully lower volatility and shallower worst-case drawdowns. You pay for that by missing the early part of every recovery.
So here is the version I actually think holds up in practice.
- Decide your full allocation to the asset first, independent of any signal.
- When the 50-day is above the 200-day and the 200-day itself is rising, run full size.
- When the 50-day is above the 200-day but the 200-day is flat or falling, the trend is young and unconfirmed, so run half.
- After a death cross, cut to a quarter position or whatever core you would hold through anything, and stop adding.
- Never execute on the day of the cross. Scale in or out over a week or two. The signal carries months of lag, so entry precision buys you nothing.
- Budget for whipsaws. In a sideways year you may get flipped two or three times, and each flip costs a little. That is the fee for being small during the disasters.
The failure mode to avoid is the one from the top of this post, treating the cross as a buy button and the chart screenshot as a forecast. Check the state of the averages once a month, adjust size when it flips, and let the unglamorous effect of being big in trending markets and small in bleeding ones accumulate. The cross will never get you the bottom or the top, and it was never supposed to.