Learn Day Trading With a Clear Process: Rules, Replay and a Useful Journal

Video: How I Would Learn Day Trading FAST (If I could start over)
Creator: Craig Percoco
Original Video: Watch the original video
Difficulty: Beginner to intermediate

The Big Idea

The quickest useful learning path is one that reduces wasted decisions. Craig Percoco argues that beginners often reverse the order: they fund an account, copy an entry idea, and only later discover that they cannot explain the rules or assess the results. His proposed alternative is to learn foundations, specialize, test a simple model, and practice execution before putting money at risk.

The central distinction is between learning a method and experiencing an exciting outcome. A profitable trade can happen by chance. A losing trade can follow the rules correctly. Neither outcome alone tells you whether your decision process improved.

This guide turns that distinction into measurable work. You will build a rule sheet, separate execution errors from model results, calculate outcomes in units of initial risk, and compare historical replay with real-time paper trading. A small structure-break strategy supplies a consistent reference for the exercises. It is an original research baseline, not a reconstruction of the presenter’s partially undisclosed trading system.

If You Remember Only One Thing

Advance when your records show what you can do consistently and what remains uncertain. A winning streak, a course completion or a persuasive chart example is not a substitute for that evidence.

60-Second Summary

• Choose learning goals you can measure: clear definitions, accurate risk calculations and complete records.
• Start with one instrument and one model to reduce the number of changing variables.
• Learn how orders, costs, chart data and position sizing work before evaluating a setup.
• Judge process quality separately from individual trade results.
• Use actual average wins and losses to assess expectancy; planned targets are not realized payoffs.
• Test additional filters one at a time against an unchanged baseline.
• Use replay to practice quickly, then paper trading to expose timing and execution problems.
• Treat the difference between stages as information, rather than assuming simulation proves live profitability.

Build Evidence Instead of Chasing a Daily Dollar Goal

Percoco recommends process goals as a way to avoid making immediate income the measure of early progress. A useful goal specifies a behavior you can inspect. “Become disciplined” is vague. “Record the entry, stop, initial risk and reason before every simulated order” leaves a clear record.

Our example: in a 20-opportunity paper session, you identify 18 correctly, execute 15, and document every opportunity. Setup recognition is 90%; execution completion is 75% of the eligible set. The account might finish positive or negative. Those process measures still identify where practice belongs. They do not establish that the setup has a positive edge.

The source sometimes describes mastery of the process as leading naturally to income. The necessary qualification is that consistent execution can faithfully reproduce a losing method. You need both execution quality and evidence that the method’s net outcomes are favorable.

A Learning Plan With Observable Milestones

The video favors a sequence over random strategy collecting. The following milestones are our practical interpretation, not a prescribed calendar or a promise of rapid competence.

These are diagnostic gates. No fixed number of trades certifies readiness. The source mentions roughly 200 trades as an example of building conviction, but 200 observations can still be unrepresentative, highly correlated or selected after seeing outcomes.

Specialize to make mistakes easier to diagnose

Percoco’s one-market emphasis reduces the beginner’s decision load. Changing the instrument, timeframe, entry model and exit method together makes it difficult to explain a result. Keeping them stable makes differences in recognition and execution easier to detect.

Specialization is a learning choice, not proof that multiple-market approaches cannot work. Pick a research instrument that fits your available practice time and whose contract or share mechanics you understand. The source’s preference for cryptocurrency reflects his circumstances; 24-hour access and leverage do not make an instrument inherently safer.

Use reading to answer concrete questions

The presenter highlights Trading in the Zone by Mark Douglas as influential to his thinking. He also names other reading material, but the supplied transcript does not reliably identify all editions or authors. Treat these as his reading suggestions rather than verified purchasing recommendations.

For any resource you study, extract a question you can apply: “How do I respond to a rule-following loss?” “What makes a level identifiable before the outcome?” “Which execution assumption is hidden in this example?” Reading supports practice when it changes a definition or a measurable behavior.

Separate a Bad Result From a Bad Decision

The video warns against revenge trading and treating every loss as personal failure. A useful journal records two judgments: whether the trade followed the defined method, and what the trade earned or lost. This prevents money from becoming the only feedback signal.

Suppose your plan forbids entry after a missed trigger. You chase anyway and earn +1.5R. Recording only the profit rewards the wrong behavior. Tagging it as a rule violation preserves the distinction. Conversely, a clean −1R stop belongs in the model’s results even if the trade felt uncomfortable.

Emotional control does not mean emotions disappear. Write a response to a recognizable trigger: after a sizing error, stop the practice session, reconcile the order and document the cause before continuing. This is an example of a process control, not a source rule. Its purpose is to prevent an error from silently changing the experiment.

The presenter strongly discourages outside pressure and favors an accountability partner. A useful partner reviews your rules, records and exclusions rather than broadcasting impressive profits. Avoiding distracting commentary need not mean hiding financial risks from people affected by your decisions. The transcript’s broad claim that announcing goals produces a dopamine response that reduces success is not supported by research supplied here; it is not needed to justify choosing constructive feedback.

Learn the Mechanics Under the Chart Labels

The source discusses market structure, volume, support and resistance before advanced confluences. Start by making each word observable. Market structure describes a sequence of highs and lows. Support and resistance describe areas where previous price behavior suggests a potential response. Neither label guarantees a future reversal.

A “break of structure” needs a precise reference level and a trigger. Is the level a prior session high, a confirmed swing high or a discretionary zone? Does a wick qualify, or must the candle close beyond it? A “change of character” generally describes a break contrary to an established sequence, but that requires an explicit definition of the preceding sequence.

Our code uses confirmed swing highs and closing breaks. It does not call every bullish break a change of character because it does not classify the prior bearish trend. That difference keeps the research label honest.

Volume also needs context: the feed and instrument determine what is measured. A simple chart is easier to inspect, but simplicity itself is not evidence of predictive power. Indicators and discretionary skills remain components to evaluate, rather than explanations for all profitable outcomes.

Calculate R and the Win Rate That Actually Breaks Even

1R is the initial planned dollar risk on a trade. If 50 shares enter at $100 with a stop at $98, initial gross risk is 50 × $2 = $100. A $200 gross gain is +2R; a $100 gross loss is −1R. Deduct $12 of costs from that winner and its net result becomes +1.88R. Actual losses can exceed −1R after gaps and adverse execution.

For a simplified system whose winners always earn bR and whose losers always lose 1R, gross expectancy is E = p × b − (1 − p). The break-even win rate is 1 / (b + 1).

The video’s 40% and 30% examples are profitable points above break-even, not the minimum thresholds. With an average 0.10R of costs, the three modeled thresholds rise to 55%, 36.67% and 27.5%. The formula becomes (1 + c) / (b + 1), where c is cost in R. These calculations assume constant payoffs and costs; use the actual conditional averages when the outcomes vary.

Partial exits explain why a headline target can mislead. In our example, closing half at +1R and half at +3R earns +2R on the original full position before costs. If the second half instead exits at the entry price, the full-position result is +0.5R. Two attractive target labels do not add into a five-R trade.

The presenter reports his own win rate and large profits, but the transcript supplies no complete reconciled trade history, capital basis or cost methodology. Those reports motivate his discussion; they do not establish expected results for the reader.

Make Replay an Honest Test of Decisions

Replay accelerates observation because you can advance bars instead of waiting through a session. Its value depends on deciding before the next price movement appears. Start from dates selected in advance, include quiet sessions, and keep future candles hidden while recording your choice.

1. Freeze a dated rule sheet and identify a consecutive historical period.
2. At each decision point, record the known level, eligibility and planned action before advancing.
3. Record skipped opportunities, signals while occupied and orders that never fill.
4. Advance price, apply the same exit rules and record costs and net R.
5. Review only after the agreed block; save any proposed revision as a new version.

The source suggests looking for commonalities among winners and losers. That is a useful way to generate hypotheses. It does not independently validate a filter selected from the same results. If losers often occurred below an RSI threshold, freeze that hypothesis and evaluate it on later data. Removing those losers retrospectively only tells you how a revised rule would fit the old sample.

A journal that connects rules to results

Keep both the rule-following model ledger and the all-actions execution ledger. Preserve mistakes in the latter; deleting them makes your practical implementation look better than it was.

Compare Replay With Real-Time Paper Trading

Percoco distinguishes quick historical practice from waiting and acting in a live market. Paper practice can reveal missed entries, premature exits and position-sizing delays. It adds a different kind of information even when no money is at risk.

Use the same frozen definitions and maintain an ideal rule-based record beside your actual paper actions. For each opportunity, ask whether the discrepancy came from recognition, delay, sizing, order behavior or discretionary interference. A change in average net R alone cannot identify the cause.

Results may deteriorate from replay to paper to live execution, as the source anticipates, but that is not a required mathematical sequence. Different periods and fill models can also change results. Paper fills do not guarantee live liquidity, and live losses are not automatically evidence of poor psychology. The method itself may have changed effectiveness.

Executable Research: Does One Confluence Improve a Simple Baseline?

//@version=6
strategy("CandleChasers: Simple Structure Learning Lab", overlay=true, initial_capital=100000, currency=currency.USD, default_qty_type=strategy.percent_of_equity, default_qty_value=10, pyramiding=0, commission_type=strategy.commission.percent, commission_value=0.05, slippage=2, margin_long=100, margin_short=100, calc_on_order_fills=false, calc_on_every_tick=false, process_orders_on_close=false)

// Independent learning baseline. Not Craig Percoco's private strategy.
// Use standard candles on a USD stock; position sizing is exposure, not risk.
pivotBars = input.int(5, "Swing confirmation bars on each side", minval=1)
atrBars = input.int(14, "ATR length", minval=1)
stopATR = input.float(2.0, "Initial stop distance in ATR", minval=0.1, step=0.1)
targetR = input.float(2.0, "Target in initial stop distances", minval=0.1, step=0.1)
useRSI = input.bool(false, "Add experimental RSI entry filter")
rsiBars = input.int(14, "RSI length", minval=2)
rsiFloor = input.float(50.0, "Minimum RSI for entry", minval=0, maxval=100)

if syminfo.type != "stock" or syminfo.currency != "USD" or syminfo.pointvalue != 1
    runtime.error("Use a USD stock with point value 1 and standard candles.")

confirmedHigh = ta.pivothigh(high, pivotBars, pivotBars)
atr = ta.atr(atrBars)
rsi = ta.rsi(close, rsiBars)
var float activeHigh = na
var bool consumed = true
var int stopTicks = na
var int targetTicks = na

// A pivot becomes usable now, never on its earlier visual swing bar.
if barstate.isconfirmed and not na(confirmedHigh)
    activeHigh := confirmedHigh
    consumed := false

breakSignal = barstate.isconfirmed and not consumed and not na(activeHigh) and close > activeHigh and close[1] <= activeHigh
closedThisBar = strategy.closedtrades > nz(strategy.closedtrades[1], 0)
eligible = breakSignal and strategy.position_size == 0 and not closedThisBar and strategy.equity > 0 and not na(atr) and atr > 0
filterPass = not useRSI or (not na(rsi) and rsi >= rsiFloor)

// Consume the first qualifying break even if filtering/position state skips it.
if breakSignal
    consumed := true

if eligible and filterPass
    stopTicks := math.max(1, int(math.ceil(atr * stopATR / syminfo.mintick)))
    targetTicks := math.max(1, int(math.round(stopTicks * targetR)))
    strategy.entry("Long", strategy.long)
    // Relative tick brackets are anchored to the eventual entry fill.
    strategy.exit("Bracket", "Long", loss=stopTicks, profit=targetTicks)

if strategy.position_size > 0
    strategy.exit("Bracket", "Long", loss=stopTicks, profit=targetTicks)

plot(activeHigh, "Latest confirmed swing high (available now)", color=color.orange, style=plot.style_stepline)
plotshape(eligible, title="Eligible first break before RSI filter", style=shape.circle, location=location.belowbar, color=color.blue, size=size.tiny)
plot(strategy.position_size > 0 ? strategy.position_avg_price - stopTicks * syminfo.mintick : na, "Stop from actual fill", color=color.red, style=plot.style_linebr)
plot(strategy.position_size > 0 ? strategy.position_avg_price + targetTicks * syminfo.mintick : na, "Target from actual fill", color=color.green, style=plot.style_linebr)

The source recommends building a simple model before adding filters. We can make that advice concrete: does adding an RSI entry threshold improve net expectancy on a fixed structure-break baseline, and what opportunities does it remove? The code tests the chart component of that question. It cannot measure a person’s learning speed or discipline.

This code tests one objective interpretation of the source’s concept. Any mechanical definitions introduced here are ours and are not necessarily the creator’s exact rules.

Source Rules vs Experimental Definitions vs Implementation Shortcuts

The video’s RSI illustration concerns exits and re-entry. Our RSI entry filter is a different, explicitly independent experiment. The source does not provide all private-model rules, indicator logic or gap criteria needed to reproduce that system.

Execution Review and Validation status

A pivot becomes usable only after the right-side confirmation bars finish. The orange line begins when it is known; it is not backdated to make an earlier entry possible. On a confirmed closing break, a market order is submitted and normally becomes eligible to fill at the next bar’s opening tick. The initial ATR distance is frozen. Relative tick stop and target orders are submitted with the entry and anchor to its eventual simulated fill.

The default position allocation is 10% of equity as exposure, not 10% equity at risk. For a $100,000 account, about $10,000 of exposure with a stop 2% below entry represents approximately $200 gross planned loss, before costs and sizing adjustments. Change allocation in Properties, and check the actual quantity and initial risk in the trade list. Instrument increments and a large gap can affect fills or affordability.

The script uses 0.05% commission per executed order and two ticks of slippage as illustrative assumptions. Replace them with justified estimates in Properties. Stop execution can be worse than the trigger price. Tick rounding makes the achieved target ratio approximate. Both exits touched on one bar require lower-timeframe inspection or a Bar Magnifier sensitivity check; do not assume the target happened first.

There is no pyramiding, reversal or same-bar re-entry after a closed trade. New pivots can replace the reference for future signals but cannot move an open trade’s frozen bracket. The first break consumes its level even if filtering or an existing position prevents entry. Older swing levels disappear when newer ones confirm. These choices reduce the opportunity set compared with tracking every historical level.

Use standard candles. This baseline has no session-close liquidation or maximum holding period and can carry positions overnight; it is a structure research exercise, not a complete day-trading execution model. There is no higher-timeframe data request. Audit a pivot’s confirmation time, entry fill, stop and target before interpreting results. TradingView’s official strategy documentation (paid link) explains the broker emulator and relevant execution settings.

Validation status: This Pine v6 strategy has been reviewed as text but has not been compiled in TradingView or run against historical data here. No performance results are claimed.

What This Code Actually Tests

The baseline enters long on the first closing cross above the newest confirmed swing high while flat, subject to sufficient indicator history and positive equity. The optional RSI filter accepts or rejects that entry. Exit orders use fixed ATR-based distance and an adjustable multiple of that distance from the actual fill. Blue markers show eligible signals before the RSI filter within that run.

Filter runs may finish positions at different times and therefore become eligible for later signals differently. Their marker sets need not match. For a genuinely common comparison denominator, maintain a separate replay list of all first breaks independent of position occupancy, and report which each variant could trade.

What This Code Does Not Test

It does not reconstruct the presenter’s private strategy, establish his reported returns, identify a full change-of-character sequence, or trade retracements into fair-value gaps. It omits shorts, RSI-based exits, scale-outs, multiple concurrent setups and personal execution errors. It does not establish that specialization or a particular learning sequence improves income.

Backtest Lab

These are our proposed experiments. Choose a development period and a later untouched evaluation period before comparing variants. Keep instrument, timeframe, dates, allocation, entry definitions and costs fixed unless that item is the variable under study.

A filter adds value only if its effect survives fresh evaluation and fits the intended objective. Report trade count, conditional average wins and losses, break-even frequency, net expectancy, profit factor, drawdown and open trades at the sample end. Treat an isolated excellent setting surrounded by poor neighboring settings as a question to investigate, not a reason to select it.

Try It Yourself: One Practice Block, Three Separate Scores

Select ten consecutive historical sessions for practice on one instrument. This is a learning exercise, not a validation-sized sample. Before revealing outcomes, record all first breaks and whether the written baseline allows entry. Maintain three scores: correct setup identification, correct execution and net R. Do not merge them into a single success grade.

Original worked example: eight closed trades produce +2R, −1R, −1R, +2R, −1R, −1R, +2R and −1R. Gross total is +1R and average is +0.125R. At 0.10R average costs per trade, net total falls to +0.20R and average to +0.025R. A small favorable result can be fragile. Neither the eight-trade sample nor perfect adherence proves an edge.

Ask a second reviewer to label several charts with your rule sheet and without outcomes. Resolve disagreements by changing the definition, then date the new version. Test the revision on a fresh block rather than silently relabeling the old sample.

Check Your Understanding

1. Can a profitable trade be an execution mistake?
Yes. An unauthorized chase can win. Record both the rule violation and the financial result.

2. What is the gross break-even win rate for +3R winners and −1R losers?
25%. Costs and variable realized payoffs change that threshold.

3. Does 10% account exposure mean 10% account risk?
No. Planned loss depends on quantity and the entry-to-stop distance; gaps and costs can increase actual loss.

4. Why cannot the code enter at the original pivot bar?
The swing is not confirmed until the specified later bars finish. Acting earlier would use future information.

5. Why retest a journal-derived filter on new data?
It was selected because it fitted known outcomes. Fresh observations assess whether that relationship persists.

6. What does paper trading add after replay?
A record of timing, waiting, recognition and order-handling behavior under real-time conditions. Its fill model still cannot guarantee live execution.

Final Takeaway

The source’s useful lesson is to make learning deliberate: specialize enough to diagnose mistakes, define a simple model, gather honest records and compare execution across practice stages. Consistency supports evaluation but does not guarantee profitability. The next useful question is whether one clearly specified change improves a frozen baseline on fresh data without hiding missed opportunities or costs.

Risk Disclosure

This material is for educational and research purposes only, not financial or investment advice. Trading can result in substantial losses. Historical and simulated outcomes do not guarantee future results. Costs, liquidity, leverage, gaps, execution and changing market conditions can materially affect outcomes and cause losses beyond planned risk.

This article is an independent educational companion to the video and is not affiliated with, sponsored by, endorsed by, or authorized by Craig Percoco.