People who look at the long/short ratio often treat it as a contrarian indicator: there are too many accounts going long, retail traders get squeezed to the side, so it should fall next. AIN last night was a counterexample. At exactly 22:00, on Binance USDT perpetuals, the global long/short accounts ratio was 3.92, with nearly 80% of accounts going long; over the next hour, AINUSDT perpetuals moved from 0.0232 to 0.0400, up 72.4%.
This line turned only after the price had already risen: this morning at 11:00 it dropped to 0.84, and the number of short accounts overtook the long accounts. It didn’t report ahead of the price and signal the turn—it followed behind the price and then flipped.
One coin doesn’t prove much. I pulled the top-of-the-hour data from the next day when trading, one by one, for the U-margined perpetual contracts, and sampled 523 out of 528; 5 contracts with Chinese names couldn’t be retrieved.
There were 98 contracts whose long/short ratio was above 3 late yesterday morning; by this morning at 11:00, 64.3% of them had risen. Across all 523 contracts, the proportion was 62.0%. In the group where longs were crowded together, that day it didn’t drop any more.
The contrarian read fails because this line mainly responds to price. I broke each contract down hour by hour, for a total of 14,641 samples: in the hour when the price rose, the share of periods where the long/short ratio fell at the same time was 56.2%; in the hour when the price fell, the share where the long/short ratio rose at the same time was 60.5%. You could use the change in the long/short ratio in this hour to predict the next hour’s price: after the ratio increased, the next hour rose in 50.3% of cases; after the ratio decreased, it rose in 53.1% of cases. The latter is only slightly more in line with the contrarian interpretation, but the linkage is far weaker than within the same hour—pretty close to a coin toss.
My inference is: whenever the price moves, some people take profit and others open positions against the trend, so the number of accounts skews toward the opposite direction. So reading the long/short ratio as contrarian mostly captures the echoes of the price itself—rather than the direction of the next move.
Two limitations: the sample is only for one day, and most of the contracts were up on that day; the long/short ratio counts accounts, not the size of positions.
$AIN #多空比 #perpetual contract
This line turned only after the price had already risen: this morning at 11:00 it dropped to 0.84, and the number of short accounts overtook the long accounts. It didn’t report ahead of the price and signal the turn—it followed behind the price and then flipped.
One coin doesn’t prove much. I pulled the top-of-the-hour data from the next day when trading, one by one, for the U-margined perpetual contracts, and sampled 523 out of 528; 5 contracts with Chinese names couldn’t be retrieved.
There were 98 contracts whose long/short ratio was above 3 late yesterday morning; by this morning at 11:00, 64.3% of them had risen. Across all 523 contracts, the proportion was 62.0%. In the group where longs were crowded together, that day it didn’t drop any more.
The contrarian read fails because this line mainly responds to price. I broke each contract down hour by hour, for a total of 14,641 samples: in the hour when the price rose, the share of periods where the long/short ratio fell at the same time was 56.2%; in the hour when the price fell, the share where the long/short ratio rose at the same time was 60.5%. You could use the change in the long/short ratio in this hour to predict the next hour’s price: after the ratio increased, the next hour rose in 50.3% of cases; after the ratio decreased, it rose in 53.1% of cases. The latter is only slightly more in line with the contrarian interpretation, but the linkage is far weaker than within the same hour—pretty close to a coin toss.
My inference is: whenever the price moves, some people take profit and others open positions against the trend, so the number of accounts skews toward the opposite direction. So reading the long/short ratio as contrarian mostly captures the echoes of the price itself—rather than the direction of the next move.
Two limitations: the sample is only for one day, and most of the contracts were up on that day; the long/short ratio counts accounts, not the size of positions.
$AIN #多空比 #perpetual contract