Reading leverage in context: structure and liquidity, together
Leverage is one of the trickiest forces to read well, because the same number can point in opposite directions depending on where price sits. We recently upgraded how Orion reads it, and the change is a good window into how the engine grows: from a clean first version to one that cross-checks itself against market structure and liquidity before it draws any conclusion.
A crowded book, on its own, is not a direction
One of the conditions Orion tracks is the long/short account ratio: the balance of traders positioned long versus short on perpetual futures, averaged across venues. The first version of this reading was deliberately simple. When a large majority of accounts were long, it treated that as downside risk, on the idea that a lot of leverage pointing one way is fuel for a sharp move if price turns against it.
That logic holds up near a top. Near a bottom it reads the opposite way. Through one accumulation phase, Ether ground higher from around $1,500 toward $2,700 while the crowd sat heavily long the whole way. The ratio was high, but the context was the opposite of a top: price was near the low of its range, and the resting liquidity that would actually move it was stacked above, not below.
A crowded long/short ratio is not a direction. It is a fact that means one thing at a suppressed low and the opposite at a stretched high.
Reading it in context
The upgrade is simple to describe. Before the engine concludes anything from a crowded book, it now gathers two more pieces of context:
- Structural position. Where price sits inside its own range, so a reading near a low is treated differently from one near a high.
- Liquidity skew. Whether the liquidation liquidity is stacked above price, leaving room to squeeze upward, or below it, leaving room to cascade downward. This is read from the live liquidation heatmap.
Only then does it lean. Crowded longs at a stretched high with liquidity below read bearish, a genuine cascade setup. Crowded longs at a suppressed low with liquidity above read bullish, the accumulation pattern where the fuel to move price sits overhead. Anything in between reads as a neutral positioning note, with no direction attached. The same cross-check now runs on the short side, and the same logic feeds the altcoin read through Ether, the alt bellwether, so the engine holds off on calling a top or a bottom early.
Cross-check both sides, always. A number that describes the crowd only becomes useful once you know where price sits in its structure and where the liquidity that will move it is resting.
An honest note on the data
The liquidity skew comes from a live liquidation heatmap, which is a strong proxy for where the fuel sits rather than a full order-book depth map. It builds up as the engine runs, so on a fresh start it has less to work with. In that case the new read stays deliberately conservative: when it cannot see the liquidity picture clearly, it holds at neutral rather than leaning. We would rather the engine say less than say something it cannot support.
Why we version the method, in public
Every method we ship keeps getting better, and we keep that evolution on the record rather than editing it away. This leverage read is now version two. Version one stays in the public track record, scored exactly as it landed, and version two starts a clean series from here, dated and labelled, so over time you can compare the two and judge the upgrade on its results.
There is one line we hold. We only set aside a stretch of history when the data capture itself was incomplete, which is a measurable, mechanical reason. We never quietly remove readings we simply like less. A track record is only worth something when it keeps the full picture, and that is exactly what lets you trust the parts of it that look good.
Where the lenses stand
For context, here is how the public, fully-scored readings break down by lens at the time of this change. Leverage version one is the outlier, and versioning it is how we bring it up to the standard the other lenses already meet.
| Lens | 30-day directional accuracy |
|---|---|
| Cycle | 100% |
| Structure | 58% |
| On-chain | 58% |
| Leverage (version one) | 40% |
We are letting the headline number stand rather than smoothing it. As version one ages out and version two accumulates, the figure moves on its own, which is the only kind of improvement worth publishing.
Everything here describes conditions and method, not advice. Orion publishes readings of the market's state; what you do with them is yours. You can browse the live readings on the analytics page, see every one scored on the track record, or start with the method overview in How Orion reads the market.