Solana on-chain signals
Where the network stands against its own history, and the one forecast this data supports. Everything here is measured, and everything that failed measurement is listed at the bottom rather than quietly dropped.
Volatility outlook
This call was made 26 July 2026 and resolves 9 August 2026. Historical hit rate at this horizon: 66%.
Volatility of SOL against BTC, so a market-wide move is not counted as a Solana event. The relationship is real but modest — it explains roughly 3% of the variation, which makes it a regime gauge and not a trading rule. It says nothing about direction; see below for why nothing here does.
State of the network
Each figure against its own history. For anything that grows over time the percentile shown is of the 30-day move, not the level — stablecoin supply sits at the top of its own range almost every day simply because the series climbs, and reporting that as a signal would be theatre.
| Metric | Now | 30 days | Percentile of that move | Reading | |
|---|---|---|---|---|---|
| Stablecoins on Solana | $17.03B | +10.9% | 78 | capital arriving on the chain | |
| DEX volume, 30 days | $52.26B | −20.7% | 27 | trading thinning out | |
| Velocity (volume ÷ stablecoins) | 3.18× | −24.2% | 26 | money parked, not trading | |
| DeFi TVL | $4.82B | +3.6% | 58 | capital committing | |
| Fees paid to the network, 30 days | $15.01M | +36.0% | 73 | users paying more for blockspace | |
| Real staking yield | 0.02% | — | 4 | staking barely outruns dilution | |
| Total staked | 428M SOL | — | — | ||
Real staking yield is the nominal rate minus dilution — what a staker gains in share of total supply, rather than in SOL. Its percentile is of the level, because it is a bounded rate rather than a growing total.
What this data cannot tell you
Direction. Fourteen indicators were tested against forward returns at five horizons; not one carried usable information. That is a result, not a gap in the collection — the same pipeline scores +0.52 correlating TVL with the return that already happened, so it detects relationships when they exist.
How this was tested — 14 indicators, 210 tests, and everything that failed
The one that survived
| Forward window | Correlation | p (corrected) | Hit rate | Independent obs. | Sign stability |
|---|---|---|---|---|---|
| 3 days | +0.141 | 0.018 | 57.1% | 484 | stable in 3/3 |
| 7 days | +0.187 | 0.043 | 62.2% | 207 | stable in 3/3 |
| 14 days | +0.225 | 0.086 | 65.8% | 103 | stable in 3/3 |
Seven attempts to break it
Cleared the bar, not yet confirmed (3)
| Indicator | Against | Window | r | p | Caveat |
|---|---|---|---|---|---|
| Velocity, 30d change | volatility | 7d | +0.168 | 0.080 | correlates +0.958 with the confirmed signal — the same thing measured twice |
| Real staking yield | drawdown | 7d | +0.255 | 0.021 | plausibly a roundabout measure of stake leaving; mechanism unclear |
| Fees to the network, 30d change | drawdown | 3d | +0.115 | 0.082 | weakest of the group, and only at the shortest window |
Failed outright (10)
| Indicator | Best against | Window | r | p | Why rejected |
|---|---|---|---|---|---|
| Self-financing, 30d | drawdown | 7d | +0.176 | 0.112 | correlation sign flips between sub-periods; p=0.112 after correction > 0.1 |
| Deactivating stake / total staked | drawdown | 14d | +0.128 | 0.373 | p=0.373 after correction > 0.1 |
| Deactivating vs its own 90d median | drawdown | 14d | +0.110 | 0.457 | p=0.457 after correction > 0.1 |
| Velocity: DEX 30d / stablecoins | direction | 7d | -0.061 | 0.930 | correlation sign flips between sub-periods; p=0.930 after correction > 0.1; hit rate 52.2% < 55% |
| Stablecoin supply, 30d change | direction | 7d | -0.055 | 0.972 | correlation sign flips between sub-periods; p=0.972 after correction > 0.1; hit rate 49.7% < 55% |
| Self-financing, 30d change | volatility | 7d | +0.118 | 0.398 | correlation sign flips between sub-periods; p=0.398 after correction > 0.1 |
| DeFi TVL, 30d change | drawdown | 14d | +0.107 | 0.459 | correlation sign flips between sub-periods; p=0.459 after correction > 0.1 |
| App fees / protocol fees | volatility | 14d | -0.353 | 0.003 | correlation sign flips between sub-periods |
| Net stake flow / total staked | volatility | 7d | -0.049 | 0.941 | p=0.941 after correction > 0.1; hit rate 51.4% < 55% |
| Deactivating / activating | volatility | 7d | +0.051 | 0.941 | p=0.941 after correction > 0.1; hit rate 51.7% < 55% |
Method
| Target | SOL against BTC, not dollars — in dollars a network metric tracks the crypto market and the market's moves get credited to Solana |
| Overlapping windows | effective sample = observations ÷ horizon, used inside every p-value. Four years of daily data is worth ~16 independent observations at 90 days |
| Multiple comparisons | Benjamini-Hochberg within each target (return: 28 / vol: 37 / dd: 37 tests) |
| Revisions | indicators lagged past each source's revision window (7d DefiLlama, 14d Blockworks) |
| Volatility control | partial correlation against previous realized volatility — otherwise anything that co-moves with turbulence appears to predict it |
| Passing bar | n_eff ≥ 100, p ≤ 0.1 corrected, hit rate ≥ 55%, sign identical in all three sub-periods |
Method v2.0, scored 2026-07-27. Data: DefiLlama (volume, stablecoins, TVL, fees, prices) and Blockworks (validator and staker revenue, stake flows), the latter also behind /solana-onchain/.