Paper Trading to Live: The Hardest Transition

Paper trading feels like a video game. Nothing is at stake, so you follow every signal without a second thought. Then you go live, the account drops ₹50,000 in a week, and a calm rule follower turns into someone checking prices at traffic lights. You skip a signal. You cut a position the model wanted to keep. The strategy did not change. You did.

The gap between paper and live is real and it has three parts: execution, data and psychology. The first two you can model. The third you can only manage, and the way to manage it is to go live in steps small enough that your nerves can keep up with your rules.

Those numbers are not about systematic investors. They are about people who went live with real money and no process. I put them first because they show what happens when the gap between a plan on paper and behaviour under pressure is left unmanaged.

Why paper trading results don't transfer directly

The execution gap

On paper, you fill instantly at the price on screen. Live, you join a queue, pay the spread and, for a large order, eat into the book. Module 10.2 put numbers on it: about ₹246 statutory cost on a ₹1 lakh delivery round trip before any slippage. A paper account that ignores this will look 1% to 3% a year better than reality for a strategy that trades often.

The data gap

Many paper setups compute signals on the closing price and assume a fill at that same close. Live, the signal is known only after the close, and you trade the next session. For a momentum strategy that buys what is already rising, that one day lag is a real cost. Either trade in the closing auction for F&O stocks or model the next day open honestly.

The psychology gap

Losses hurt roughly twice as much as equal gains feel good. That is one of the most replicated findings in behavioural finance, and you will feel it the first time a real position is down 12%. On paper, you hold. Live, you look for a reason to sell. Path 8 covers the biases in depth; here the point is practical: plan for them before the money is real.

The right way to paper trade

Model realistic costs

Charge your paper account every cost in Module 10.2: STT, stamp duty, exchange and SEBI fees, depository charges, GST, brokerage, plus a slippage assumption by liquidity bucket. A sensible start is 0.05% for the 100 largest stocks, 0.15% for the next 150 and 0.3% or more beyond that.

Use live data at the time you would really trade

Record the price at the minute your process would place the order, not the day's close. If your process is "rebalance at 10 am on Monday", your paper fill is the 10 am price on Monday.

Write down every decision

Signal, price, what you did, and if you did anything other than what the rule said, why. This journal is the most valuable output of the paper phase, because it shows you where you will be tempted to override.

Minimum paper duration: 100 trades or 3 months, whichever is longer

Twenty trades tell you nothing. Twenty trades in a rising market tell you less. Aim for at least 100 trades and at least one real correction inside the window. The year to 9 October 2026 had one: the Nifty 50 fell from its record close of 26,329 on 2 January to 22,520, about 15%. A paper record that never saw a fall like that has not been tested.

The transition protocol: 5 phases

This is the staircase I use. Each step has an entry test, and the steps exist to stop me from overriding myself at the wrong moment.

1. Paper, 3 monthsEvery cost charged, 100 trades, one correction included.
2. Live at 10%Same signals, same timing, real fills.
3. CompareAfter 30 live trades, within 2 points a year of paper.
4. Scale to 50%At least one quarter, journal shows no overrides.
5. Full sizeOnly after the 50% stage behaved like the plan.

Worked example: ₹10 lakh intended capital

Nine months sounds slow. It is cheap insurance. If the first live month brings a 10% fall, it costs you ₹10,000 instead of ₹1 lakh, and you are far more likely to stick with the rules.

The psychology of real money

Loss aversion and the override temptation

The rule says exit. The position is right at the exit level. You think "one more hour". That hour is the override, and every override turns a tested system into an untested one. Write the rule down in plain words before you go live, and when the moment comes, read it. Pre-commitment beats willpower.

Revenge trading

A loss, then a bigger next trade to win it back, then a bigger loss. Hard cap the loss per day and per week. When the cap hits, you stop for the day. No exceptions, including for the trade that "looks perfect".

Drawdown tolerance shock

A 15% fall on paper is a number. A 15% fall live is your child's school fees. Decide the largest drawdown you can live through before you start, compare it with the strategy's back tested maximum drawdown, and if your number is smaller, run less capital. That is a sizing decision, not a character test.

Common mistakes in the transition

• Going 100% live on day one. The first bad week then hits at full size, the rules break, and the second week is worse.
• Changing the strategy after three losing trades. A healthy systematic strategy can be right only 40% to 55% of the time. Three losses in a row are ordinary. Judge after 30 trades, not three.
• Forgetting T+1. Shares bought today are not in your demat until settlement. A plan that sells them tomorrow before that creates a delivery problem.
• Keeping backtest slippage after you have live data. Once you know your real fills, use them in sizing.
• Paper trading only in a rising market. Include at least one correction.

Measuring success: when is your strategy working?

• At least 50 to 100 live trades before any verdict.
• Sharpe ratio, live against paper: within about 0.3 is normal noise.
• Maximum drawdown, live against back tested: live worse by more than half again is a warning.
• Average slippage against your cost model: recalibrate if it is double.
• Overrides in the journal: the number that should be zero.

Decision tree

• Live Sharpe within 0.3 of paper, drawdown in range, slippage as modelled: continue.
• Sharpe 0.3 to 0.5 lower, drawdown somewhat deeper: adjust sizing or execution timing.
• Sharpe more than 0.5 lower, drawdown far deeper: stop and re-examine the signal or the risk rules.

How this shows up in RupeeCase

RupeeCase is built around the same three steps: compare, validate the back test, invest. Every index page shows its model, back tested history, maximum drawdown and full rebalance record before you put in a rupee, so you can do your own paper phase by following the rebalances for a few cycles. When you do invest, you can start with a fraction of what you plan to commit and add later, because the shares sit in your own demat. The track record page and an index like Nifty 50 Equal Weight are good places to start watching.

Questions people ask me

What to remember

• Paper misses three things: execution costs, timing and your own nerves.
• Paper trade with every cost charged, for 3 months and 100 trades at least.
• Go live at 10%, compare, scale to 50%, then full size.
• Write rules down before you need them; read them when you want to break them.
• Judge after 50 to 100 live trades, not after three.