Onchain PvE / PvP Index
A 0–100 daily score of whether onchain memecoin markets are PvE (net new money, broad participation, most tokens rising together) or PvP (flat or shrinking money, gains concentrated in a few names, bots extracting value). Solana first; Base, BNB Chain, Ethereum and Robinhood Chain are included where data exists.
1. The index
Token universe
About 5,000 memecoin and launchpad tokens across five chains, discovered from CoinGecko categories, launchpads and DEX listings, then cleaned for duplicates and obvious junk. On a given day a token counts only if it has at least 50 days of price history (this removes launch noise), and a day is scored only when at least 20 tokens are eligible. Single-day price jumps above 10× are treated as data errors: the following 14 days are dropped for that token.
Five components
| Component | Weight | What it measures | Raw signals |
|---|---|---|---|
| Breadth | 40% | How many tokens are rising | Average of four shares of eligible tokens: above their 20-day average, above their 50-day average, 30-day return beating the chain's base asset (SOL, BNB or ETH), positive 7-day return. |
| Capital | 20% | Is new money arriving | 30-day change in stablecoin supply on Solana and across all chains; SOL vs BTC over 30 days. |
| Volume | 20% | Is activity expanding | 7-day vs 90-day average of Solana DEX volume, all-chain DEX volume, and pump.fun revenue. |
| Concentration | 10% | Are gains spread out (PvP tell, inverted) | Share of all positive 30-day returns captured by the top 5 tokens (returns capped at +2,000%). |
| Extraction | 10% | How much bots take (PvP tell, inverted) | Jito MEV tips per dollar of Solana DEX volume, 7-day sum, in basis points. |
From signals to a score
- Breadth is used as an absolute 0–100 share.
- Every other raw signal is converted to a rolling 180-day percentile: the share of the past 180 days with a value at or below today's (at least 90 days required). Only past data is used, so there is no lookahead. Capital and volume average their signals' percentiles.
- Concentration and extraction are inverted (100 − percentile), so a high value is always PvE-like.
- The score is the weighted average of the five components. If a component is missing for a day, the remaining weights are rescaled. A day is only scored when breadth exists and at least 60% of the weight is available.
- The result is smoothed with a 3-day average.
Thresholds
PvE ≥ 65, PvP ≤ 40, neutral in between. They were chosen on in-sample data only (April 2023 – June 2025); everything from July 2025 on is out-of-sample.
The equal-weight variant (all five components at 20%) is shown as a robustness check: when it agrees with the main score, the reading does not hinge on the weights.
2. Data cleaning
The routine runs about an hour after the UTC day closes. DeFiLlama's first print of that day is often missing slow-reporting DEXes, so the latest day can read about 40% too low. Two safeguards handle this:
- Outlier clipping (full history): an isolated 1–2 day spike from a single DEX is replaced by that DEX's trailing 7-day median.
- Provisional fill (last 2 days only): a DEX that reported on at least 5 of the prior 7 days but is missing, partial or stuck on a repeated value is topped up to its 7-day median. Fills are recomputed on every run, so real data replaces them as soon as it lands. Older days are never altered, so genuine drops are not hidden.
The dashboard marks provisional days (hatched orange bars) and the run log lists every substitution.
3. Episodes
An episode is a run of days with score ≥ 65, merging runs separated by 7 days or less. Entry is the close of the day after the episode starts; exit is the close of the day after the last PvE day. Episodes starting from July 2025 are out-of-sample.
4. Runners
For each 2026 episode, tokens are ranked by return from entry. Only tradable tokens are ranked: estimated entry market cap ≥ $1M (current supply × entry price), excluding tokens with single-day jumps above 10× or disagreeing price sources.
- Peak x: highest price held for at least 2 days within 30 days after the episode ended, divided by the entry price.
- Hold return: return when holding until the index exits PvE, or until 14 days after the episode ends.
- Tokens launched during the episode are flagged.
5. Strategy backtest
Rule-based baskets are bought at episode entry and sold at exit, then compounded across episodes:
- A: equal weight across all eligible tokens. B / B2: top 10 by 7-day / 30-day momentum. B3: B limited to market cap ≥ $1M. C: top 10 by estimated market cap. E: top 10 revivals. +trail30: the same with a 30% trailing stop. D: SOL only.
- Costs: 2% round trip per token trade, 0.5% for SOL. Each token's gain is capped at +300%.
- Random baseline: the same rule and hold length started on random non-PvE days. A rule matters only if it clearly beats its random counterpart.
6. Big-cap breadth
A re-implementation of the "Breadth crypto" app, recomputed from Binance data.
- Universe: the top 150 coins by market cap with a Binance USDT spot pair, excluding stablecoins, wrapped, staked and leveraged tokens. Each day's list is locked once saved, so history never changes retroactively.
- % above MA200: share of members whose daily close is above their 200-day simple average (needs 200 closes). % above MA50 uses 50 closes. At least 20 coins are required.
- The MA20/50/100 lines are averages of the %MA200 line; RSI 14 (Wilder) is computed on both breadth lines. Readings below 20 are flagged as extreme.
- The current, still-open daily candle is reported separately and never stored as final.
7. Trench Index
The official daily score comes from tehcscreener.com. It weights four activity components: launchpad volume 35%, bot volume 25%, tokens launched 20%, bot fees 20%. Each is a percentile vs the trailing 90 and 365 days, blended 60/40. Rises apply the same day; falls ease in (new = raw + 0.65 × (previous − raw)). Bands: Ded 0–19, Cold 20–39, Active 40–59, Hot 60–79, Euphoric 80–100.
The dotted proxy line is our own recomputation from the public per-launchpad and per-bot series. Those series cover a narrower set than the official components, so the proxy is a cross-check, not a replacement. Exact component values are snapshotted daily going forward.
8. Solana DEX volume
Daily Solana spot DEX volume from DeFiLlama (UTC days), cleaned as in section 2. 20/50/100-day simple averages are overlaid; the lower panel is RSI 14 (Wilder) of daily volume, not of SOL price.
9. Sources & schedule
All inputs are free and public: DeFiLlama (DEX volume, fees, stablecoins, token prices), Binance (daily closes), CoinGecko (universe and categories), GeckoTerminal (pool prices and volume), DexScreener, Jupiter, pump.fun, tehcscreener.com and Solana RPC.
The dashboard is rebuilt twice a day at about 08:05 and 20:05 ICT (UTC+7). Because the index is daily, the evening run usually shows the same score date as the morning run.
10. Limitations
- Thresholds were tuned on 2023-04 to 2025-06; judge the signal on out-of-sample episodes.
- The token universe is built from today's discovery sources, so very old dead tokens may be under-represented (survivorship bias).
- Free APIs rate-limit. When a host keeps refusing requests, cached data is used for that run and the log records it.
- The newest one to two days can be provisional and are revised automatically.
- Market caps in the runners and backtest tables are estimates based on current supply.
- This is analysis, not investment advice.