May 12, 2026 8 min read

Paper-first vs live-first — the operating model behind sustainable quant betting

Live-first is the default everywhere. It's also why most quant betting programs fail. Paper-first methodology — proving the strategy before risking capital — is the discipline that separates 6-month operators from 6-year ones.

Paper Trading
Methodology

Most quant betting programs go live-first: the operator writes a strategy, places real bets immediately, watches P/L drift, and either gets lucky enough to validate a real edge or unlucky enough to blow up before they can tell the difference. Paper-first reverses the order: every strategy runs paper-only against live odds until the operator promotes it based on CLV evidence. The discipline is what separates 6-month operators from 6-year ones. Glitch Edge enforces paper-first in code — the platform refuses live placement on un-promoted strategies.

Why live-first fails

The math is brutal. Three independent failure modes compound:

Variance hides truth. A strategy with no edge can show +20% ROI for 100 bets due to variance. A strategy with a real +2% edge can show -15% ROI for the same period. You cannot tell the difference from inside the run.
Operator psychology breaks under drawdown. Most operators tap out during a 30%+ drawdown — emotionally, financially, or because their bankroll is now too small to size correctly. The strategy “fails” not because the edge was wrong but because the human couldn’t hold position.
Model errors compound silently. A model overestimating edge by 30% sizes bets correctly for the estimated edge and dangerously for the actual edge. Live-first means discovering this with real money.

Each of these is survivable individually. All three together, with a small bankroll, are how most programs fail in their first 6 months.

What paper-first actually does

Paper-first means every new strategy runs against live odds without placing real money. The platform records what the bet would have been, what it would have paid, and what the closing line looked like — but no capital changes hands.

This gives you four things you can’t get from live-first:

Validation of model accuracy. If paper P/L drifts from expected over 200+ bets, your model probability estimates are wrong. Discover this on paper, not in live.
CLV signal at zero capital risk. CLV is the cleanest short-term signal of edge. Paper-first lets you measure CLV before betting.
Execution-path validation. The strategy can actually pull live odds, compute fair prices, and decide to bet on the cadence you expect. Bugs surface in paper without losing money.
Drawdown profile preview. If the paper drawdown is 3× what your model predicts, the strategy has a real problem — broken assumptions, not just unlucky.

The promotion rule that works

Paper-first only matters if there’s a real promotion step from paper to live. Without it, paper runs are an academic exercise. The rule that works in practice:

A strategy is promotion-eligible when it meets all three:

50+ paper bets recorded. Below this, CLV signal is too noisy.
Positive aggregate CLV (typically +0.5% or better).
Drawdown profile within model expectation (max paper drawdown ≤ 1.5× model-predicted max drawdown over the same period).

The operator must explicitly promote — no auto-graduation. Glitch Edge surfaces promotion-eligible strategies in the weekly report but requires manual approval.

What promotion is NOT

Three things promotion does not mean:

Promotion is not a profit guarantee. Even a promoted strategy can drawdown 25% before recovering. Paper validates edge probability, not specific outcomes.
Promotion is not all-or-nothing. Recipes within a strategy can be promoted independently. A cricket strategy might have its “audience refresh” recipe live while its “creative fatigue swap” stays paper.
Promotion is not permanent. A live strategy that drawdowns past its threshold gets halted. After halt, it stays halted until the operator un-halts (typically after model review + re-promotion).

The friction is the point

Operators new to paper-first complain about the friction: “I want to test this strategy now, not in 50 bets.” The friction is the discipline. The first 50 bets in paper teach you whether the strategy works; the strategies that get promoted are the ones that survive the friction.

A second-order benefit: paper-first eliminates the “press during a hot streak” reflex. By the time a strategy is promoted, you’ve seen its variance pattern in paper — including the hot streaks that aren’t predictive — and the temptation to press during a live hot streak is dampened.

What paper-first is NOT solving

The honest framing matters here. Paper-first is one discipline lever, not the only one. It doesn’t fix:

Model decay. Markets get more efficient over time. A strategy with a real edge today may have decayed in 6 months. Paper-first doesn’t catch decay; rolling CLV does.
Correlated exposure. If you have five strategies that all bet the same NBA game in the same direction, paper-first doesn’t help — they were each paper-validated independently. Per-bet bankroll caps + correlated-exposure caps handle this.
Bookmaker limiting. A strategy with positive CLV that’s getting account-limited isn’t sustainable. Paper-first doesn’t surface this; the operator has to watch for it.

These need separate guardrails. Paper-first is necessary but not sufficient.

What this looks like operationally

A typical Glitch Edge operating week:

Day	Activity
Monday	Review last week’s paper + live P/L + CLV. Flag strategies for promotion or halt.
Tue–Wed	Author new strategies (or refine existing). Strategies start paper-only by default.
Thursday	Refresh existing live strategies (re-tune Kelly fraction, update brand-config).
Friday	Outreach hour — citation tracker review, content updates, customer questions.

The platform doesn’t add work to your week; it adds structure. The work was always going to happen — paper-first means it happens in the right order.

Pricing models that work for paper-first

Glitch Edge specifically: Free tier is paper-only, Pro tier is $29/mo with live placement. The Free tier exists because of paper-first methodology — most users should be paper for their first month regardless of subscription, so charging for paper would be perverse.

For managed-service operators running paper-first for clients: typical pricing is a setup fee ($500–$2,000) that covers strategy authoring + first-30-day paper run + the promotion decision, then a monthly retainer ($300–$800) that covers ongoing operation + monthly strategy review.

Frequently asked questions

How long does paper mode take to validate a strategy?

50 bets is the floor for any signal at all. 200 bets is where CLV signal becomes reliable. 500+ bets is where drawdown profile becomes interpretable. Depending on bet cadence, that’s 2–8 weeks for most strategies.

Can I skip paper mode if I have a strategy I’m confident in?

Glitch Edge doesn’t let you. The platform refuses live placement on un-promoted strategies. The friction is the point; confident operators tap out at the rate confident-but-wrong operators do.

What if my strategy doesn’t pass promotion criteria?

Then it doesn’t get promoted. Most strategies don’t. Treat this as the platform doing its job — you saved capital that would have gone to a strategy that wasn’t ready.

Does paper P/L map to live P/L exactly?

Close, with two caveats: paper doesn’t model book limits (a winning paper account doesn’t get limited), and paper doesn’t model placement latency (paper assumes the price you saw is the price you got). For most strategies, paper P/L is within 5% of live P/L over comparable samples.

Can I run multiple strategies in paper simultaneously?

Yes — Free tier supports one strategy in paper, Pro tier unlimited. Many operators run 5–10 strategies in paper concurrently and promote the 1–2 that earn it.