quantbase.live
Backtest checking for crypto · runs inside Claude or Cursor · by avasis.ai

Stop funding luck you mistook for skill.

Everyone else sells you the backtester. This is the referee: a Sharpe deflated against a trial count the server keeps and you cannot edit, costs pinned at the exchange's real fee, and a holdout that opens once, ever. Ninety-four tools whose job is to take your best result away from you.

PerspectiveTrader

You still write the strategy and you still run the backtest. The only thing you lose is the ability to forget the ones that failed.

Twenty-four trials · random tradesthe tallest bar is the one you would have funded
Every bar is a strategy with no edge. One of them is tall. That is not a signal; it is what twenty-four attempts at nothing look like once you keep the best.
Servermcp.quantbase.live · one endpoint
DataCoinbase spot · hourly since 2015
Refereedeflated Sharpe · PBO · reality check · holdout
Connect · under a minuteClaude Desktop · Cursor · Claude Code · any MCP client
{ "mcpServers": { "quantbase": {
    "url": "https://mcp.quantbase.live/mcp",
    "headers": { "Authorization": "Bearer <your key>" } } } }
Paste it in, ask your agent to run gate_check. The key is free at sign-up; the ledger starts counting from your first trial.
94,090
BTC-USD hourly bars
94 / 5
Tools / irreversible
27
Assets in top30-usd
0.2.0
Server version
Is this you?

What you have, and what you get back.

You have

A strategy you are tuning — in Python, in a notebook, in an agent.

You get

A Sharpe deflated against every variant you tried — counted by the server, not by you. Plus PBO, a confidence interval, and a holdout that opens once.

Needs Claude, Cursor or any MCP client.

You have

A backtest someone else is showing you — a signal group, a bot seller, a pitch.

You get

Their Sharpe, deflated against how many things they tried. Type two numbers and watch what is left of it.

Check one now → No account.

You have

An idea and no data. You want to know if it survives before you fund it.

You get

The server brings the prices and the exchange fees. You bring the idea. 200 trials a month, free.

No card. Email only.

Not for you if you are picking coins to buy, or want signals. This server emits none — it only tells you whether a result you already have is real.

01 — The server

One endpoint. Ninety-four tools.

Your client talks to one server. The server owns the data, the splits, the statistics and the trial count, and reaches out to the exchange for prices and fees.

CLAUDE · ANY MCP CLIENT
QUANTBASE MCP quantbase.live/mcp · 94 tools
DATA8 toolsIngest, catalog, quality report, gap policy, activity bars
Coinbase · prices
RESEARCH DESIGN5 toolsImmutable splits, registration, freeze, forks, the ledger
TEST EXECUTION6 toolsBacktests, cross-section, sweeps, execution replay
THE GAUNTLET7 toolsDeflated Sharpe, PBO, Reality Check, SPA, M_eff, bootstrap
COST REALISM2 toolsFees pinned server-side. No tool takes your assumption
Coinbase · fees
GATING2 toolsTen checks, then one holdout — enforced by a unique index
KNOWLEDGE BASE5 tools496 documented strategies, 40 blowup case studies
LIVE TRADING5 toolsArm, disarm, orders, paper sessions — frozen and passed only
Coinbase · orders
ACCOUNT3 toolsStatus, plans, limits
Solid rule · tools inside the server Amber rule · outbound to the exchange
02 — What it is for

Luck+

Thirty curvesrandom trades · no edge in any of them

Luck has a shape. You have seen it before.

Thirty equity curves, every trade in them random. The best one is simply the top of the fan, and it is the one you would have funded.

Thirty strategies with no edge at allthe best of thirty in amber
Fig. 1 Same data, same odds, different luck. The winner would pass any "Sharpe above 1.5" filter you care to set. Its only edge is that it was chosen.
30
Strategies, none of them with an edge
1
Looks like genius: the top of the fan
0
Information in the ranking. It is noise, sorted.
Limit · Selection bias

Any test you re-run until it passes will pass. That includes significance tests and Monte Carlo reshuffles: they measure the winner and cannot see the losers you threw away.

PerspectiveQuant

A Sharpe of 1.5 sounds like a grade. It is a measurement, and it means nothing until you know how many times you measured.

03 — Why it happens

Hurdle+

The one formulaE[max SR] ≈ σ·√(2 ln N)

The bar moves right with every attempt.

A Sharpe ratio is only good relative to the best of N tries. Try thirty variations, keep the winner, and you have cleared a much lower bar than you think. N is the whole correction, so the server keeps its own append-only count. You cannot pass it in, lower it, or delete a trial.

The hurdle rises with every attemptsimulated · no edge present
Fig. 2 Filled bars sit past the hurdle. The hurdle itself moves right as N grows, so a result has to clear the best that chance would produce, not zero.
2.1σ
The bar after ten tries, in units of noise
3.0σ
After a hundred
0
Ways to lower N: the count is append-only
Theorem · The bar rises with N

The best of N attempts with no edge is expected to score about σ·√(2 ln N). Everything the server computes starts from this line: the deflated Sharpe is the chance your result beat it.

PerspectiveQuant

N is the entire correction, which is exactly why it is the one number you are not allowed to type in.

04 — The finding

Finding+

btc-ma28 training runs · then ten assets, no refit

Same rules. Ten assets instead of one.

A moving-average family on BTC returned a Sharpe of 0.95 in-sample, the best of 28 recorded training runs. Run unchanged across ten assets, the pooled figure is negative. The gap between those two numbers is the product.

0.95Sharpe
One asset · in-sample · BTC-USD · best of 28 training runs
−1.376Sharpe
Ten assets · pooled · identical parameters, no refit
The same rule on ten assets3 of 10 positive
+++−−−−−−−
Fig. 3 The assets that traded less scored better. The ordering tracks turnover cost, not signal: a rule that looked like an edge on one asset was mostly a fee schedule.
1.19
Best asset · AERO-USD
−0.61
Rank correlation, Sharpe against trade count
28
Training runs, counted by the server
Result · Move the rule before you fund it

Breadth is the cheapest honesty test there is. A rule with a real edge survives being moved; a rule fitted to one asset's history does not, and no amount of in-sample Sharpe tells you which you have until you move it.

PerspectiveData engineer

Ten assets on one exchange are not ten tests. The next section is about how few they really are.

05 — Breadth

Breadth+

27 × 27correlation · ordered by cluster

Twenty-seven assets. Three real tests.

Crypto moves together. Ordered by cluster, the correlation matrix has three bright blocks and little else, so twenty-seven assets buy you three independent looks, not twenty-seven.

Pairwise correlationbright = moves together
Fig. 4 Three blocks, not twenty-seven dots. The server counts effective tests from this, not from the length of the asset list.
0.52
Mean pairwise corr.
57%
Variance in PC1
3.0
Effective tests
06 — Cost

Cost+

Coinbase taker feepublished · non-promotional entry tier

Sixty basis points, every side.

Costs are checked before a run, not after. Most strategies die here, and it is cheaper to find out now.

What one trade costslog scale on the last bar
Per side
60bps
Round trip
120bps
20 trades / day, annualised
8,760%
Fig. 5 The fees alone cost more per year than most strategies ever make. A rule that only works with zero fees is not a strategy; it is a description of the fee.
Assumption · What is measured and what is drawn

The matrix is ordered by cluster and drawn to illustrate the structure; the 0.52, 57% and 3.0 are measured on the server's own data. The fee is Coinbase's published taker rate at the entry tier, which is what a new account actually pays.

PerspectiveTrader

Buy the fee schedule before you buy the idea. The round trip decides how much edge a strategy can ever keep.

07 — The gate

Gate+

btc-matrain split · v0.2.0

A Sharpe never appears on its own.

It arrives with the trial count that produced it, the deflated figure, and the confidence interval, drawn at the same weight when that interval straddles zero as when it does not.

0.95
Sharpe, reported
28
Trials, server count
unopened
Holdout
straddles 0
Deflated Sharpe
95% interval, after correction for 28 trials✕ fails gate
0
Fig. 6 The interval crosses zero, so the result is consistent with no edge at all. The point estimate sits inside it; the ink line is zero.
Limit · No second look

The server will not hide an interval that crosses zero, and it will not let a strategy be re-tested until the trial count says so. There is no route from a failed gate to the holdout.

PerspectiveTrader

The holdout is the moment the backtest stops being yours and becomes evidence.

Research budget

Your history can only pay for so many attempts.

A fixed span of data supports a maximum number of honest tries. Past it, no result can be told apart from the best of many coincidences. These are this server's own families, right now — including the one that has already spent more than its data can cover.

FamilySpentHistoryBudget
btc-ma29 trials2.16y heldoverspent
btc-tsmom16 trials8.67y held60 left
xsec-donchian9 trials9.33y held97 left
ev-capitulation7 trials10.33y held168 left
btc-faber4 trials8.67y held72 left
08 — Pricing

Pricing+

Metered by trialsholdouts separate · free needs no card

Priced by trials, because trials are the cost.

Free needs no card. Holdouts are metered separately: each one opens exactly once per strategy, ever.

Free
$0
Trials / month200
Sweep points100
Universe10
Holdouts / month1

Enough to take one honest idea through the whole gauntlet.

Start free
Pro
$49 / month
Trials / month5,000
Sweep points1,000
Universe50
Holdouts / month10

For running a family properly: sweeps, splits, ten-asset pooling.

Choose Pro
Team
$399 / month
Trials / month50,000
Sweep points5,000
Universe200
Holdouts / month100

Shared ledger across a desk, with per-tenant isolation.

Choose Team
Definition · A trial

One backtest, one sweep point, or one cross-section run. Every trial permanently raises the bar your own results must clear, so it is the scarce thing worth pricing, and a smaller sweep buys you a lower hurdle as well as a smaller bill.

PerspectiveTrader

The referee says no far more often than yes. That is the product, not a flaw in it.