Real money study: does money rushing into a chain mark the top?
Version 1, 4 Oct 2026, 00:10 Bangkok. Built for the risk on/off dashboard. All numbers below are from free public data (DefiLlama, Robinhood Chain Blockscout). Latest version always at this same link.
Read this first (plain words)
- Solana: when stablecoins pour into Solana fast (+8% or more in 14 days, 7 day average), the chance of another 30% crash in meme coins over the next 3 weeks is about 3 times normal (34 to 44% vs 9 to 15%). This held in every one of 45 test versions.
- For pump.fun coins the average next 3 weeks after an inflow was negative in every year (2024, 2025, 2026): about -17% to -21% vs flat.
- For older Solana memes (BONK, WIF type) it depended on the year: in the 2024 bull run inflows were fuel (+3% to +32%), in 2025 they were the top (-21% to -30%). So treat it as a danger flag, not a sure sell.
- Robinhood Chain (only one cycle so far): stablecoin inflow peaked the same week as the meme top (+$195M in the week to 6 Sep), then stopped (about $0 after 13 Sep). Meme coins fell 56% from there.
- New wallets on Robinhood peaked the week AFTER the top. They lag the money. Money flow is the earlier, better signal.
- Now (2 Oct): Solana stablecoins +2.8% in 14 days (7 day average): normal, no flag. Robinhood: flat for 3 weeks: no rush.
1. What "real money" means here
- Stablecoin supply on the chain (USDC, USDT, etc, in dollars). New dollars arriving = buyers bridging in. Source:
https://stablecoins.llama.fi/stablecoincharts/Solanaand.../Robinhood, daily. - DEX volume (how busy trading is). Source:
https://api.llama.fi/overview/dexs/Solana?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=true. - New and active wallets. Robinhood: Blockscout charts (
robinhoodchain.blockscout.com/stats-service/api/v1/lines/newAccounts, works only from a real browser, Cloudflare blocks servers). Solana: no free daily history found (DefiLlama active users only start 23 Aug 2026; bridges data is paid; no Dune key). - Targets: point in time meme baskets (dead coins included): Solana older memes (Feb 2024 to Sep 2026), pump.fun coins (Dec 2024 to Sep 2026), bonk.fun coins, Robinhood memes (Aug to Oct 2026).
2. Solana: stablecoin inflow vs what happened next
Rule tested: stablecoin supply (7 day average) up more than X% over the last N days. Outcome: basket change over the next 21 days and the chance it fell 30%+ at some point inside those 21 days.
| Rule | Basket | Days flagged | Next 21d avg (flag vs rest) | Chance of -30% (flag vs rest) | By year (flag days) |
|---|---|---|---|---|---|
| +6% in 14d | Older memes | 253 | -8% vs +8% | 37% vs 14% | 2024 +4%, 2025 -29%, 2026 +7% |
| +8% in 14d | Older memes | 190 | -12% vs +7% | 41% vs 15% | 2024 +3%, 2025 -30%, 2026 +2% |
| +8% in 30d | Older memes | 344 | +5% vs +3% | 34% vs 12% | 2024 +26%, 2025 -21%, 2026 -4% |
| +6% in 14d | pump.fun | 134 | -17% vs -1% | 35% vs 11% | 2024 -29%, 2025 -21%, 2026 +3% |
| +8% in 14d | pump.fun | 105 | -21% vs -1% | 44% vs 11% | 2024 -29%, 2025 -23%, 2026 -2% |
| +12% in 30d | pump.fun | 128 | -21% vs 0% | 44% vs 10% | 2024 -24%, 2025 -20%, 2026 -28% |
- Robustness: 45 versions (3 baskets x windows 14/30/45 days x thresholds 4 to 12%). Crash odds higher after inflows in 93% of versions. Average return worse in 71%.
- Bigger and faster inflow = worse. +12% in 14 days: pump.fun -30% next 21 days on average, crash odds 58%.
- Bonk.fun basket: weaker and mixed (short history, 2025 to 2026).
- Honest limit: flagged days come in clusters (about 25 separate inflow episodes since Feb 2024). Looking only at the first day of each episode, 2024 episodes were mostly up (bull run fuel), 2025 to 2026 episodes mostly down (median about -10%). Read the flag as "late stage, crash risk up", strongest for pump.fun coins.
- Outflows (stablecoins falling) were NOT a clean bottom signal.
At the actual turns
- 24 tops and bottoms in the Solana memes basket (30% swings, Feb 2024 to Sep 2026): stablecoin flow showed no repeatable pattern exactly at the turn (median +11% over 30 days at tops vs +9% at bottoms).
- DEX activity did: 7 day volume vs 60 day average about 1.25x at tops and 0.77 to 0.82x at bottoms. But low activity alone is not a buy: days with volume under 0.8x of normal were followed by -11% (older memes) on average. Quiet is where bottoms happen, not a signal that one is in.
3. Robinhood Chain: one full cycle (Jul to Oct 2026)
| Week to | Meme basket | Stablecoins in ($M) | New wallets per day | Active wallets per day | DEX volume per day |
|---|---|---|---|---|---|
| 2 Aug | 1.2 | +54 | 104k | 263k | $436M |
| 9 Aug | 2.5 | +48 | 109k | 275k | $454M |
| 16 Aug | 2.5 | +55 | 974k (one off spike, 11 to 12 Aug) | 1.37M | $481M |
| 23 Aug | 2.7 | +75 | 94k | 385k | $537M |
| 30 Aug | 3.2 | +56 | 115k | 350k | $967M |
| 6 Sep (top) | 3.2 | +195 | 144k | 394k | $2.60B |
| 13 Sep | 2.8 | +109 | 173k | 458k | $2.02B |
| 20 Sep | 2.5 | -4 | 143k | 425k | $1.58B |
| 27 Sep | 1.8 | -2 | 132k | 392k | $1.38B |
| 4 Oct | 1.4 | +17 | 114k | 326k | $1.44B |
- Money rush and trading peak = the top week. New and active wallets peaked one week later. After the money stopped, memes fell 56%.
- Note from an earlier session (27 Sep): most Robinhood stablecoins (about $855M of $1.07B) sit in Morpho savings (Robinhood Earn USDG, Ethena loops), not waiting to buy memes. So the useful part is the weekly CHANGE, not the level.
- One cycle only: this is a strong story, not yet a proven rule.
4. New wallets: worth paying for a source?
- What the data says: on Robinhood, new wallets peaked a week after the top and after the money. They confirm the top late. They do not warn early.
- Money flow (stablecoins in) and trading volume already catch the rush at the top week. New wallets mostly repeat that a week later, so they are largely redundant for timing.
- Recommendation: do not buy a data source now. If you want it anyway for Solana, a paid source (Dune, Artemis, Allium, about $100 to $400 a month) would let us test whether Solana new wallets lead. Expected value is low given the Robinhood result.
- Free middle path, already available: DefiLlama daily active users per app (pump.fun, PumpSwap, Raydium, fomo Wallet) since 23 Aug 2026. Record it daily from now; test after the next Solana cycle. Endpoint:
https://api.llama.fi/overview/active-users(use totalDataChartBreakdown, per app).
5. How to plug this into the risk on/off dashboard
Inputs (daily)
- S = stablecoin supply on the chain, 7 day average (DefiLlama stablecoincharts).
- V = DEX volume 7 day average divided by 60 day average.
- MONEY RUSH (danger): S up 8%+ over 14 days. Action word: SELL SOME on chain memes (pump.fun type especially). Expect crash odds about 3x normal for 3 weeks.
- MONEY RUSH EXTREME: S up 12%+ over 14 days. Action: SELL most chain meme exposure.
- MONEY STOPPED (Robinhood pattern, testing): weekly inflow drops from a big week to about zero while price is near the high. Action: SELL SOME.
- No flag: HOLD as per the other dashboard rules. Outflows and quiet volume are not buy signals on their own.
6. Readings now (2 Oct 2026)
- Solana: stablecoins $16.5B, +2.8% in 14 days (7 day average): no flag. DEX volume 1.01x normal.
- Robinhood: stablecoins $1.08B, about flat since 13 Sep; DEX volume 1.04x of its 60 day average: no rush.
7. Files behind this (for another session)
- Cowork session scratchpad macro/money/: m1.py (state tests), grid.pkl (45 versions), sol_daily.pkl, stable_*.json, tvl_*.json, dex_*.json, fees_*.json, rh_blockscout.json (Robinhood daily new/active wallets and transactions, 1 Jun to 3 Oct 2026), active_users_bd.pkl.
- Basket lines: http://159.223.88.168/hx-c9ois/edge/chain-baskets-2026-10-01/ (idx_cmc.json inside all.tgz).
Version 1, 4 Oct 2026 00:10 Bangkok. Latest version: http://159.223.88.168/hx-c9ois/research/real-money.html