How Paper Trading Results Differ From Live

Paper trading results almost always look better than live trading results, and the gap is not random noise. Every assumption a simulator makes when it cannot know your real queue position, your market impact, or the price movement during order routing happens to favor the paper trader. The result is a consistent upward bias that stacks across every trade rather than canceling out over time. Understanding the paper trading vs live results gap before you commit capital is the only way to set a realistic expectation for what your strategy will actually do.

This article breaks down each simulator assumption individually, explains which types of strategies are most exposed to each one, and shows you how to measure your own gap by running both environments simultaneously. The goal is not to discourage paper trading but to help you read its output correctly.

The Gap Is Not Random

Why simulators skew optimistic

Every assumption a simulator makes that differs from live trading leans in the same direction: fills are easier to obtain, sizes are accepted without question, and latency disappears. This is not a coincidence or a flaw in any particular platform; it is an inherent property of any simulation that cannot observe your actual queue position, your order's market impact, or the price movement that occurs while a live order is being routed. The consequence is not noise around true performance but a consistent upward bias that compounds across every trade you take.

Understanding each assumption individually matters because the size of your personal gap depends on your specific strategy. A market-order strategy in a large-cap stock will have a different gap profile than a limit-order strategy in a thinly traded name. Naming each assumption lets you predict which ones apply to you before you go live rather than discovering them through losses.

Paper trading is still a useful tool

None of this means paper trading is useless. Simulators catch code bugs, misconfigured webhooks, and order-routing logic errors that would cost real money to discover. The goal of paper trading is not to preview exact profit and loss but to verify that signals fire, orders reach the broker, and position state updates correctly after each fill.

Alpaca's own documentation is direct on this point: even small amounts of real money can often provide insight into issues not seen in a simulation environment.¹ Paper trading handles the logic layer; real money handles everything else.

Assumption One: Reference-Price Fills

What simulators fill at

Alpaca's paper environment matches all orders against the best available current market price, the NBBO, rather than routing to an exchange.¹ TradersPost paper accounts fill market orders at the midpoint between the bid and the ask by default. That midpoint is a price no real market participant can guarantee on a marketable order, and it quietly credits you with half the spread on every trade, a benefit that does not exist in live execution.

Wider spreads amplify this effect. In a thinly traded symbol where the bid-ask spread is $0.20, the midpoint assumption credits you $0.10 per share per trade. In a liquid large-cap where the spread is $0.01, the same assumption costs you almost nothing. The midpoint credit is invisible until you look for it, which is precisely why it inflates paper results without traders noticing.

How to configure the fill price assumption

TradersPost paper accounts expose a Market Price Type setting that gives you control over which quote price drives market order fills. The default is the bid-ask midpoint, but you can switch to ask-for-buys and bid-for-sells, always-ask, always-bid, or last price. Switching to ask-for-buys and bid-for-sells removes the midpoint credit and more closely approximates what a marketable order actually costs in a liquid name. For strategies that trade wide-spread or thinly traded instruments, this adjustment is worth making before drawing any conclusions from paper results.

Assumption Two: Queue Position Ignored

How a real limit order book works

When price touches your limit level in a live market, you are filled only after every order that arrived at that price before you is satisfied. A simulator fills you the moment price touches your level, treating your order as if it were first in queue regardless of when it arrived. Strategies that rely on resting limit orders, such as mean-reversion entries or scaled exits, are the most sensitive to this assumption because queue position determines whether they fill at all, not just at what price.

A limit order that paper fills instantly may not have filled live if price only briefly touched the level before reversing. The book works through queue depth in order of arrival time; if the price bounced before reaching your position in the queue, your order remains open in live trading while the paper account already shows a clean fill and a closed position.

Alpaca's explicit disclosure

Alpaca explicitly lists order queue position for non-marketable limit orders as one of the factors paper trading does not account for.¹ These gaps show up as phantom fills: trades the paper account reports as executed that would have remained open or missed entirely in live trading. The effect is not random. It consistently inflates trade count and win rate for any strategy that relies on resting limit orders.

Assumption Three: Unlimited Liquidity

Size is not checked against the book

Alpaca's paper environment does not check order quantity against NBBO quantities, meaning you can submit and receive a fill for an order much larger than the actual available liquidity.¹ In live trading, an order larger than available size at a level will be partially filled at that level, with the remainder filled at progressively worse prices, increasing average cost. A paper account running a large-size strategy in a thinly traded name will show fills that the live market could not have absorbed at the reported price.

This makes paper results for large-size or thinly traded strategies particularly unreliable as a predictor of live performance. The strategy's edge may depend on entering and exiting at specific prices; if the live market cannot absorb that size at those prices, the edge does not exist in practice regardless of how well it appeared in paper.

Partial fills in simulation

Alpaca's paper environment does introduce randomized partial fills: when an order is eligible to fill, there is a 10% chance it receives a partial fill for a random size, with the remainder re-evaluated for a subsequent fill.¹ This approximates the experience of partial fills without accurately modeling market impact. Alpaca also explicitly lists market impact and information leakage of your orders as factors not accounted for in paper trading.¹ Both of these costs are real in live trading and grow larger as your order size increases relative to typical volume in the symbol.

Assumption Four: Fills Are Always Complete

Live orders that never fill

In live trading, a non-marketable limit order may never fill if price never returns to that level, leaving the position unentered while the paper account would show a flat entry. Alpaca's documentation notes that in live trading, orders may not be filled, prices may spike, or network disconnections may require retry logic, none of which affect a paper simulation.¹ A strategy's paper win rate will look better than live in part because every triggered signal results in a clean entry, while live trading includes missed entries that vanish from the trade log entirely.

Missed entries are not neutral. They often represent the best setups in the sample, the ones where price moved so quickly it never returned to the limit level. Excluding them from the paper trade log makes the win rate look higher than live results will support.

Regulatory fees and dividends missing

Alpaca's paper environment does not simulate regulatory fees, which reduce net profit and loss on every live transaction.¹ Paper accounts on Alpaca also do not simulate dividends, so a long equity strategy that benefits from dividend income will show lower returns in paper than a live account holding the same positions.¹ These omissions are small per trade but accumulate meaningfully over a high-frequency or long-duration strategy. Regulatory fees subtract from every live transaction; dividends add to every live holding period. Depending on strategy style, the two can cut in opposite directions relative to paper results.

Assumption Five: Zero Latency

Where real latency comes from

In a live automated trading setup, latency accumulates across four phases: signal transmission, order planning, approval, and broker execution. For TradingView-sourced alerts, signal transmission alone typically accounts for 800ms to 1.5 seconds before the webhook even reaches the routing layer.² Order planning and execution add roughly 300ms to 500ms in typical conditions, making total signal-to-fill latency around 1.1 to 2 seconds under normal conditions.² A paper simulator fills at the reference price at the moment the signal fires, skipping all of this elapsed time.

Latency changes your actual fill price

During the 1 to 2 seconds between signal generation and live order placement, price can move meaningfully in a fast market, producing slippage that never appears in paper results. Including a time field in your webhook payload, formatted as an ISO-8601 timestamp, allows TradersPost to calculate elapsed time from signal generation to execution and surface latency-driven slippage in the trade signal logs.² Strategies that trade around scheduled events or volatile opens are the most exposed because price movement during that window is largest.

Assumption Six: Behavior Stays Consistent

Execution discipline in live trading

Paper trading removes psychological pressure entirely. No real loss is on the line, so there is no temptation to override a signal, exit early, or skip an entry that feels wrong. In live trading, discretionary interference with automated signals is one of the most common sources of divergence from paper results, independent of any simulator assumption. A system that looked excellent on paper but was regularly second-guessed in live trading will underperform for behavioral reasons that no fill-price adjustment can fix.

This is a variable that is genuinely difficult to measure in advance. The best proxy is a short period of live trading at reduced size, where real money is on the line but the downside is bounded. Behavior under small real-money conditions is a better predictor of behavior at full size than any amount of paper trading.

Data feed differences between environments

Different paper trading platforms use different market data sources, which affects the prices at which signals trigger and fills are calculated. Alpaca paper-only accounts are entitled only to IEX market data; live accounts have access to additional data sources, which can produce different reference prices for the same moment in time.¹ Tradier's paper environment uses data that is 15 minutes delayed, meaning signals fire and fills are calculated against prices that are already stale relative to the live market.³ These data differences introduce a layer of divergence that is unrelated to strategy logic and easy to overlook when comparing paper and live results.

Measuring Your Own Gap

Run both environments simultaneously

The most direct way to measure your paper-to-live gap is to run the same strategy in both environments at the same time and compare fill prices signal by signal. Compare the price field sent in the webhook payload, which represents the market price at alert time, against the actual fill price reported by each environment to isolate slippage. Track how many signals resulted in fills in each environment: the difference in trade count exposes missed entries that the paper account concealed.

TradersPost logs each signal alongside the price field from the webhook payload and the actual broker fill price, so you can review slippage trade by trade from the signal logs without exporting data manually. A few weeks of parallel running typically produces enough data to characterize the gap quantitatively rather than estimating it.

What a realistic gap looks like

For a liquid large-cap equity strategy using market orders, the gap is primarily latency slippage and the loss of the midpoint fill credit: small per trade but consistent in direction. For a limit-order strategy in a medium-liquidity name, queue position is the dominant variable. The paper account will show a higher fill rate than live, inflating trade count and win rate together. Sizing the gap quantitatively before committing full capital lets you set a realistic live expectation and decide whether the strategy remains viable after accounting for real-world execution costs.

If the strategy survives the adjustment, you have a more durable expectation for live performance. If it does not, you have saved the cost of discovering that in a live account.

Bottom Line

• Every simulator assumption that differs from live trading favors the paper trader; the biases stack in the same direction rather than canceling out.
• Midpoint fills, ignored queue position, and unchecked order size are the three assumptions with the largest effect on strategies that trade limit orders or low-liquidity symbols.
• Latency of 1 to 2 seconds between signal generation and live fill is real and produces slippage that paper results never show.
• Data feed differences between paper and live environments add a layer of divergence unrelated to strategy logic.
• Run both environments simultaneously, compare fill price to signal price trade by trade, and measure the gap before committing full capital.

Conclusion

The paper-to-live gap is not a reason to skip paper trading; it is a reason to read paper results with the right lens. Use the simulation to confirm that signals fire, orders route correctly, and position state updates as designed. Then measure the gap by running both environments in parallel before scaling up. Each assumption covered here, fill price, queue position, liquidity, completeness, latency, data feed, and execution discipline, is addressable once you know it applies to your strategy. The traders who close the gap fastest are the ones who name it first.

Frequently Asked Questions

Why do paper trading results almost always look better than live?

Every simulator assumption favors the paper trader: fills at the midpoint rather than the spread, no queue position for limit orders, no size check against available liquidity, and no latency. These assumptions do not cancel each other out; they stack in the same direction, producing a consistent upward bias rather than random noise around true performance. The gap is an inherent property of any simulation that cannot know your queue position, your market impact, or the price movement that occurs during live order routing.

Does Alpaca's paper trading check order size against available liquidity?

No. Alpaca's paper environment does not check order quantity against NBBO quantities, so you can receive a fill for a size much larger than actual available liquidity.¹ In live trading, an oversized order would be filled across multiple price levels, increasing average cost and reducing actual returns relative to paper. This makes paper results for large-size or thinly traded strategies particularly unreliable as a predictor of live performance.

What is the best fill price setting for paper trading?

The bid-ask midpoint default is the most optimistic setting because it credits you with a price no live market participant receives on a marketable order. Switching to ask-for-buys and bid-for-sells removes that credit and more closely approximates what a marketable order actually costs in a liquid name. For strategies that trade wide-spread or thinly traded instruments, any midpoint-based assumption will still significantly understate real execution costs.

How much latency should I expect between a TradingView alert and a live fill?

TradingView alerts typically take 800ms to 1.5 seconds to reach a webhook endpoint, before any order planning or execution begins.² Order planning and execution add roughly 300ms to 500ms in typical conditions, making total signal-to-fill latency around 1.1 to 2 seconds under normal conditions.² A paper simulator fills at the price the moment the signal fires, so it never incurs any of this elapsed time or the price movement that occurs during it.

Is it worth trading small real money instead of paper trading?

Alpaca's own documentation notes that even small amounts of real money often provide insight into issues not seen in a simulation environment.¹ Real-money trading exposes queue position effects, partial fill behavior, regulatory fees, and latency slippage that paper trading cannot replicate. It also introduces live execution discipline: the willingness to follow signals without override, which is a behavioral variable that paper trading cannot test.

References

¹ Alpaca Paper Trading Documentation
² TradersPost Signal Speed
³ TradersPost Tradier Integration