Elephant Index · Open source
The Banks Know.
Now You Do Too.
Your money has a two-tails problem: hold cash and inflation melts it; invest to escape that and an overheated market can crash. The Elephant Index positions for both tails — and predicts neither.
The problem
Two tails risk. Nowhere to hide.
Hold cash and inflation melts it — so you move into the market to keep your value. But the market runs hot: stocks and bonds at multi-decade valuations at once, an everything-bubble in Jeremy Grantham's phrase. It can crash — in 2022 they fell together and 60/40 stopped hedging — and a crash lasts: the 2000 dot-com bust took the Nasdaq fifteen years to make a new high.
It isn't bad luck — it's arithmetic and history. US government debt has passed 120% of GDP, and debt that heavy has only two ways to clear: print it away, and your cash melts; or let asset prices fall until they do, and the market crashes. Ray Dalio's study of debt crises across centuries keeps returning to that same pair of exits — with no safe place to simply wait it out. The two tails aren't a coincidence — they're the only two ways out.
So you're stuck. Just like the record $8.3 trillion parked in money funds caught in the same deadlock.
So what do you do?
The solution
Follow the tide.
The deadlock breaks the moment you stop guessing which tail fires and follow the liquidity that drives both. When it flows in, you stay invested and outgrow the melt. When it drains, you step to cash and the crash happens without you. One rule reads the tide; you check once a month.
Elephant Index = ½ · cause (liquidity) + ½ · effect (trend)
The cause sensor reads liquidity: net liquidity — the Fed's balance sheet minus the cash parked at the Fed — plus the yield curve. The effect sensor reads the trend: the asset's 10-month price trend. That sensor isn't ours — it's Meb Faber's published 10-month trend rule (2007), a public, independently replicated rule that has held up in live use for nearly two decades; trend-timing the Nasdaq itself predates even that (Siegel). Each sensor is on or off, so the Index holds 100%, 50%, or 0% — once a month. The exact parameters and code are on verify it.
An illustration of the principle — not a backtest. The Index takes the first leg down with the market and re-enters a little late; it sits out most of the crash, not all of it. A shallower hole, paid for in calm markets.
The proof
One index. Four asset classes. Six decades.
One recipe, no per-asset tuning — roughly market return at a fraction of the drawdown on equities, credit, and crypto; on gold it fails, and we publish that too. Nasdaq and Bitcoin are the live products: the highest-beta assets to the liquidity tide, where the discipline pays most.
Nasdaq
2003+
−50% → −23%
15% → 14%
Bitcoin
2016+
−75% → −50%
63% → 86%
S&P 500
1962+
−53% → −20%
7% → 8%
Dow
2003+
−49% → −23%
8% → 7%
High-yield credit
2007+
−30% → −12%
5% → 5%
Goldrejected
2003+
−39% → −38%
11% → 10%
The gold row is the published failure: the Index barely improves the drawdown and gives up a point of return, so gold was rejected. A recipe you can trust shows where it fails.
The elephant in the room
The edge is the central bank itself.
Just follow it.
Every quant pattern eventually crowds out — edges disappear once they're found, factors erode as capital piles in. The central bank is different. It is a mandated, price-insensitive actor: it must add or drain liquidity when the system demands it, no matter how many funds are already positioned. No crowd can crowd that out.
So we read it — with two sensors. The cause sensor reads liquidity itself: the central bank's net liquidity and the yield curve. The effect sensor reads price: the market's 10-month trend. The Index reflects liquidity; it doesn't predict it. When liquidity agrees with a price break, we act faster. That's the edge — and it doesn't decay.
Rules frozen since 2026, and the index held up in proxy data back to 1962 — including the 1973–74 stagflation bear in real terms (buy-and-hold lost 46% of purchasing power; the index ≈9%). Full breakdown → verify it.
