Tune Your Pocket Option Robot: Timeframe, Risk, Filters

Set Your Risk Before You Touch Any Indicator

Almost everyone tunes this backwards. They spend an evening on RSI thresholds, leave the stake on whatever the bot shipped with, and get wiped out by a losing streak that a correct entry rule would not have prevented anyway. Indicator settings change which trades you take. Stake settings change whether you survive the ones you get wrong. Set the second layer first.

How Much to Risk Per Trade

Set your risk per trade at 1–2% of account balance, and treat that as a ceiling on the entire recovery sequence, not on the first click.

That distinction is where most blown accounts start. If your trade stake is 2% and martingale is off, 2% is what a loss costs you. If martingale is on with two recovery steps at 2.25×, a single losing sequence costs roughly 8.3× your base stake — so a "2%" setting is really a 16.6% bet. Size the base stake so the whole ladder fits inside your risk budget, then work backwards to the number you type in the box.

Two more inputs decide the real damage: the asset's payout percentage and how many positions the bot can hold at once. Binary payouts are asymmetric — you risk 100% of the stake to win a fraction of it — so the payout sets the break-even win rate your settings have to clear before anything else matters. At an 80% payout you need to win about 55.6% of trades just to stand still. At 70% you need 58.8%.

The calculator below ties all of it together. Set your balance, your risk percentage, your asset's payout, and your martingale plan, and it shows you the base stake, the worst-case cost of one full ladder, how much of your daily cap that ladder eats, and the win rate you have to beat.

Setting a Max Drawdown and Trade-Count Cap

A stake rule limits one loss. A cap limits a bad day. Four settings do the work, and every serious bot exposes some version of them:

• Daily loss cap. Set a daily loss limit of 10% of balance (5% if you are conservative). Most panels label this a daily stop-loss target.
• Daily profit stop. The mirror setting, and the one everybody leaves blank: a daily take-profit at roughly the same size as your loss cap. A bot has no idea it is on a hot streak, and conditions change faster than a running configuration does.
• Consecutive-loss stop. Halt the session after 4 losses in a row. Runs of consecutive losses are how a normal win rate turns into an abnormal drawdown, and they cluster — losses in trending markets arrive back-to-back, not evenly spaced.
• Concurrent trades. Cap it at 2. Bots that scan a watchlist will happily open EUR/USD, GBP/USD and EUR/GBP at the same moment; that is not three trades, it is one euro bet at triple size. A max open trades limit is the cheapest correlation control you have.
• Trades per day. Cap at 10 on M5, 20 on M1. Without it, a choppy session lets a loose threshold generate forty entries and turn your maximum drawdown into a straight line down.

Then add a cooldown period of 15–30 minutes after the consecutive-loss stop fires. Without it, most bots resume on the very next signal — inside the same conditions that produced the streak.

Which Timeframe and Expiry Fit Your Strategy?

These two settings get confused constantly, and confusing them is why a bot with correct indicator settings still loses. The chart timeframe is the candle length the bot analyses. The expiry time is how long the trade stays open. They are independent fields, and if the expiry is shorter than the move your indicator predicted, you can be right about direction and still lose.

The working rule: start expiry at roughly twice the chart timeframe, never below one candle. A signal generated from M5 candles is a statement about the next five to fifteen minutes; giving it a 60-second expiry is asking the market to be right immediately.

Note what changes down that table besides the numbers: trade frequency collapses. If your bot fires forty times a day, no amount of martingale discipline saves you from a bad edge — you are simply paying the payout spread forty times. Slower timeframes are the cheapest accuracy upgrade available, and they cost nothing but patience.

Two filters belong in this same conversation because they gate when the bot may act at all. A trading session filter restricted to the London and New York hours removes the thin, directionless stretches where reversion signals fire and then drift. A news filter that blocks entries 15 minutes either side of high-impact releases removes the candles that no indicator setting can survive. Turn both on before you touch a single indicator period.

RSI Bot Settings for Pocket Option

The Relative Strength Index is the most common engine in Pocket Option bots because it is easy to express as a rule: below a level, buy; above a level, sell. That simplicity is also its failure mode — an unfiltered RSI rule sells every strong uptrend the whole way up.

Starting block:

• Period: 14. Leave it. Shortening to 7–9 makes the bot fire far more often on M1, and almost all of the extra entries are false signals. Change period only after you have tried changing thresholds.
• Thresholds: 25 / 75 on M5, 30 / 70 on M1, 20 / 80 on M15. Faster candles are noisier, so they touch extremes more often; the compensation is a looser threshold on fast charts and a stricter one on slow charts, not the reverse. If your bot is not trading at all, this is the first number to move — 5 points at a time.
• Direction: RSI crossing back up through the lower threshold → Call. Crossing back down through the upper → Put. The cross-back matters. Entering the moment RSI dips below 30 puts you in front of the move; waiting for it to re-cross puts you behind a turn that has started.
• Confirmation: require candle close. Intrabar RSI values move as the candle forms, so a bot that acts mid-candle acts on a number that may not exist when the candle closes. One setting, and it removes a whole class of phantom entries.
• Expiry: 2× the chart timeframe. Reversion moves are quick; give them two candles and no more.

The setting nobody documents and everybody needs: a trend gate. RSI reversion only makes sense in a range. Add a rule that blocks entries when a 50-period moving average is sloping steeply, or restrict the bot to assets that have been ranging in the session. Without it, your bot will find its worst losing streak the first time a currency pair actually trends.

MACD Bot Settings for Pocket Option

MACD is the other default engine, and it behaves almost oppositely to RSI: it lags, it wants trend, and it punishes short expiries.

Starting block:

• Periods: 12 / 26 / 9. These are the defaults for a reason — they have been the reference settings since long before binary platforms existed. On M1 only, 8 / 17 / 9 is a defensible faster variant; expect noticeably more whipsaw in exchange for more entries.
• Entry: signal-line cross, confirmed on candle close. Same confirmation logic as RSI, same reason.
• Zero-line rule. Take bullish crosses that happen above the zero line and bearish crosses below it, and skip the rest. Crosses on the opposite side of zero are counter-trend attempts — the setting that most reliably improves a MACD bot is simply refusing half its signals.
• Histogram threshold. If your bot exposes a minimum histogram value, set it just above zero. It filters the flat-market crosses where MACD line and signal line braid together and fire five times in ten candles.
• Expiry: 2–3× the chart timeframe. MACD is slower than RSI, so it needs more room. This is the single most common misconfiguration in MACD bots — correct crossover logic wired to a 60-second expiry.

Martingale Settings: Progression, Caps, and When to Turn It Off

This is the section every other page skips, and it is the setting most likely to end an account. If your bot has a martingale strategy toggle, three numbers behind it decide everything.

The multiplier is not 2

Nearly every bot defaults the multiplier to 2.0, which is inherited from coin-flip gambling where a win pays even money. Binary options do not pay even money. If your payout is 80%, a doubled stake does not recover the previous loss — it recovers 80% of it, and you drift down while "winning" your recovery trades.

The recovery multiplier you actually want is roughly (1 + payout) ÷ payout:

• 90% payout → 2.11
• 80% payout → 2.25
• 70% payout → 2.43

Notice the direction: the worse the payout, the steeper the ladder has to climb, which is exactly why martingale on low-payout assets escalates so violently.

Cap the steps, then cap the money

Set martingale steps to 2 — a maximum of three trades in any sequence. At a 2.25 multiplier that ladder costs 1 + 2.25 + 5.06 = 8.31× your base stake. Five steps costs over 50× base, which is not a recovery plan; it is a bet that you will not see six losses in a row in a market where losses cluster.

Then enforce two hard rules:

1. The full ladder must fit inside your daily loss cap. If it does not, cut the base stake until it does — never raise the cap to fit the ladder.
2. After the last step loses, reset to base and enter cooldown. A bot that resets and immediately re-enters the same conditions simply starts ladder number two.

The chart below runs one identical sequence of twenty trades — same signals, same order, an 11-win / 9-loss run at an 80% payout — through a flat 1% stake and through a capped three-step martingale. The win rate is identical. The outcome is not.

The martingale line is ahead for the first six trades. That is the trap — it works right up until the streak that it exists to survive, and then it gives back a quarter of the account in four trades. Run your own multiplier and step count through the binary martingale exposure calculator before you enable it.

When to turn it off

Turn martingale off, or leave it off, when any of these are true: your bot trades a trending market (streaks cluster), your assets pay below 75% (the ladder steepens), your daily cap cannot absorb a full sequence, or you cannot state the worst-case ladder cost in dollars without opening a calculator. If the appeal is stake variation rather than recovery, use the inverse: an anti-martingale increases size after wins, so a losing streak shrinks your exposure instead of compounding it.

Fixed-Stake Settings for Consistent Position Sizing

Fixed stake is the least exciting configuration and the one most likely to still exist in three months. Every trade risks the same amount, so nothing compounds against you.

Starting block:

• Stake: 1–2% of balance, recalculated weekly or after a ±20% balance change — not after every trade. Per-trade recalculation shrinks your stake exactly during a drawdown, which quietly makes recovery slower than the drawdown was.
• Martingale: off. These two settings are alternatives, not companions.
• Trades per day: capped harder than a martingale bot — 10 on M5, 20 on M1. There is no recovery mechanism here, so frequency is your only remaining exposure lever.
• Everything else unchanged from your indicator block above.

Fixed stake buys you something no other configuration does: a readable result. Because every trade weighs the same, the historical win rate your log reports is the actual win rate of your entry rule. Under martingale, one recovered sequence and one blown sequence weigh completely differently, and the win rate stops telling you anything about the settings you are trying to evaluate. Test on fixed stake even if you intend to run martingale later.

Trend-Following Bot Settings

If your bot supports a trend-following mode, it is usually the calmest configuration available — fewer trades, longer expiries, less time spent watching.

Starting block:

• Trend definition: EMA 12 and EMA 26, or 20 and 50 on slower charts. Direction is set by which sits above the other; strength by the slope.
• Timeframe: M15. Trend logic on M1 is mostly noise measurement.
• Entry: pullback, not breakout. Wait for price to return toward the faster EMA and resume, rather than entering on the extension. Breakout entries into a binary expiry frequently buy the exhaustion candle.
• Expiry: 2–3× the chart timeframe — 30 to 45 minutes on M15.
• Direction lock. Take Call trades only while the trend is up and Put trades only while it is down. Trend bots that also take counter-trend call/put entries are running two contradictory strategies through one set of caps.
• Session filter: on. A trend bot outside liquid hours is a range bot with the wrong settings.

Do AI Bots Like Stockley Need Different Settings?

Branded and AI-labelled tools — Stockley-style assistants and similar — present a much shorter settings panel, which readers often take to mean there is nothing left to configure. There is; the levers just moved. This is generic guidance across that whole category, not a review or endorsement of any specific tool.

Typical pocket option stockley bot settings come down to three fields: which asset (or "let the AI choose"), which time window, and which market category. Map them back onto the framework above:

• "Let the AI choose the asset" is an asset filter you switched off. Keep a whitelist. An engine that rotates onto whatever looks active can put you on a low-payout instrument where your break-even win rate quietly jumps four points.
• "Trading time" is your session filter. Same rule as before: liquid hours, and blocked around high-impact news.
• "Market category" changes the payout, and the payout changes everything downstream. Comparing a stock-index instrument against an OTC market asset is really comparing two different break-even win rates. Weekend OTC automation has its own quirks that deserve their own treatment.
• Stake and daily cap stay yours. Most of these tools return a signal you place yourself, which means expiry is your decision too — and it must match the timeframe the engine analysed, not your impatience.

The honest summary: an AI layer can replace your indicator settings. It cannot replace your risk settings, and any tool that asks you to hand those over as well has removed the only controls that were protecting you.

One Worked Configuration, End to End

Isolated numbers are easy to argue with. Here is one complete block for a $500 account running RSI reversion on M5, written the way most panels lay their fields out, with the arithmetic that makes each line follow from the one above it.

# Illustrative starting block — $500 balance, 80% payout asset
# Test on demo first. These are starting values, not results.

[account]
balance                 = 500.00
payout                  = 80          # break-even win rate = 55.6%

[risk]
risk_budget_per_seq     = 2.0         # % of balance for a FULL sequence
base_stake              = 1.20        # = (500 * 2%) / 8.31x ladder
daily_loss_cap          = 50.00       # 10% of balance
stop_after_losses       = 4
cooldown_minutes        = 20
max_concurrent_trades   = 2
max_trades_per_day      = 10

[martingale]
enabled                 = true
multiplier              = 2.25        # (1 + 0.80) / 0.80
steps                   = 2           # ladder = 1 + 2.25 + 5.06 = 8.31x
reset_on_cap            = true        # reset to base, then cooldown

[strategy]
indicator               = RSI
period                  = 14
threshold_low           = 25
threshold_high          = 75
entry                   = cross_back   # not first touch
confirm_on_candle_close = true
trend_gate_ma           = 50           # block entries on steep slope

[timing]
chart_timeframe         = M5
expiry_minutes          = 10           # 2x the candle
session_filter          = london,newyork
news_blackout_minutes   = 15

Read it top to bottom and every line is forced by the one above. The payout sets the break-even win rate and the martingale multiplier. The multiplier and step count set the 8.31× ladder. The ladder and the 2% risk budget set the base stake at $1.20 — which looks absurdly small until you notice that one full losing sequence costs $10, and the daily cap of $50 allows exactly five of them before the bot shuts itself off. That is the whole point: the small number at the top is what makes the caps at the bottom real.

If you are building a bot from scratch rather than configuring an off-the-shelf one, the same block is your parameter list — the fields change names, not meanings.

Matching a Settings Profile to Your Risk Tolerance

The same entry logic needs different numbers depending on how much variance you are willing to sit through. Pick a column and take the whole column — mixing an aggressive stake with conservative thresholds gives you the drawdown of one and the trade frequency of the other.

If you cannot decide, start in the conservative column. It generates few enough trades that you can actually read the log, and every setting in it can be loosened later without redesigning the block.

Test Your Settings on Demo Before Going Live

A settings block is a hypothesis. The only cheap way to test it is a demo account running the exact configuration you intend to fund — same assets, same session window, same stake percentages. This is a forward test, and it is worth more than any backtest a bot vendor shows you, because it runs against the spreads, payouts and execution you will actually get.

The mistake to avoid is stopping too early. A block that looks brilliant after twelve trades is telling you nothing: at that sample size, a coin flip produces "brilliant" results roughly a fifth of the time. Worse, tuning thresholds until a short run looks good is textbook curve-fitting — you end up with settings optimised for a week that has already happened.

Change one setting at a time between test runs. Changing three at once and seeing improvement tells you nothing about which one helped — and one of the other two is probably hurting.

Checking Your Settings Against Real Market Conditions

There is one gap a demo run leaves open. Demo tells you what your thresholds did; it does not easily tell you what a well-tuned threshold would have done at the same moments. When your RSI bot skips a setup, you want to know whether it was correctly filtering noise or incorrectly sitting out a clean entry.

The practical check is a side-by-side. Our free binary options live signals feed publishes live entries built on the same technical basis your bot is using — momentum, moving-average direction, trend state — so you can watch a real signal print and ask whether your current thresholds would have fired at that moment. If our feed is calling entries on an asset your bot has been silent on all session, your thresholds are too strict. If your bot has taken six trades in a stretch where the feed called none, they are too loose.

That is the whole extent of it: a reference point for calibration, not a control panel. You cannot edit your bot's settings from a signal feed, and this is not a way to make a badly configured bot behave. If what you actually want is for your bot to be fed by alerts rather than by its own indicators, that is a different setup — the Pocket Option Telegram signals channel or an MT5 Pocket Option connector route signals into a platform, whereas everything in this article assumes the bot is generating its own.

What to Change First When It Isn't Working

Most people respond to a bad session by changing everything, which destroys the only information they had. Diagnose by symptom instead, and change one thing.

• The bot barely trades. Loosen thresholds by 5 points, or drop one timeframe step — 30/70 instead of 20/80, M5 instead of M15. Do not raise the stake to make the few trades "count"; that is how a quiet week becomes an expensive one.
• The bot trades constantly and loses. Do not touch the indicator. Add confirmation on candle close, then the trend gate, then the session filter. Roughly speaking, over-trading is a filter problem, not a threshold problem.
• Losses arrive in clusters. Your entry rule may be fine; your recovery plan is not. Reduce martingale steps, or switch to fixed stake and re-measure. Clustered losses are the specific failure mode martingale cannot absorb.
• Win rate looks healthy but the balance falls. Check the payouts you actually traded. A 52% win rate is profitable at a 95% payout and losing at 80% — that is an asset-selection fix, not a settings fix. Make sure a kill switch exists and that you know how to hit it while you work this out.

And keep the settings you are running written down somewhere outside the bot. The single most common reason a configuration cannot be debugged is that nobody remembers what it was two weeks ago.

FAQ

What expiry time should a Pocket Option bot use?

Start at roughly twice your chart timeframe and never go below one candle: 1–2 minutes on M1, around 10 minutes on M5, 30 minutes on M15. Expiry is a separate field from the chart timeframe in nearly every bot, and setting it shorter than the move your indicator is predicting is the most common reason a bot with correct entry logic still loses.

Should I enable martingale on my bot?

Only if you can state the full sequence cost in dollars and that number fits inside your daily loss cap. If you do enable it, cap the steps at 2 and set the multiplier from your payout — about 2.25 at an 80% payout, not the 2.0 most bots ship with. Turn it off entirely on trending markets, on assets paying under 75%, and while you are still measuring whether the entry rule works at all.

What RSI settings work best on a 1-minute chart?

Keep the period at 14 and loosen the thresholds to 30/70, because fast candles reach extremes far more often than slow ones. Require confirmation on candle close, add a trend filter so the bot stops selling into strong uptrends, and expect to cap trades per day around 20 — an M1 RSI bot without a frequency cap will find forty entries in a choppy session.

How much should a bot stake per trade?

One to two percent of balance for the entire sequence, not for the first click. With martingale off, that is your stake. With two martingale steps at 2.25×, the full ladder costs 8.31× the base stake, so a $500 account at a 2% budget should enter a base stake near $1.20 — small enough that a complete losing sequence costs $10 rather than $83.

Do OTC assets need different bot settings?

The settings framework is the same, but the payout usually differs, and payout drives the break-even win rate and the correct martingale multiplier — so re-derive both rather than reusing a block tuned on a weekday instrument. Weekend and OTC automation has enough of its own behaviour to be worth treating separately from this configuration guide.

How long should I test settings on demo?

Judge by trade count, not by days: at least 50 closed trades, and enough calendar time to include a quiet session and a volatile one. Twelve trades is noise. Change one setting between runs so you can attribute any improvement, and record the longest losing streak alongside the win rate — that streak is what your caps actually have to survive.

Sources & Further Reading

Want to go deeper? These independent, authoritative sources shaped this guide — each one is worth reading in full:

• TradingView Help Center — Relative Strength Index (RSI): formula, overbought/oversold levels, and input definitions
• TradingView Help Center — MACD indicator: formula and fast/slow/signal length definitions
• Martingale (betting system) — why doubling after a loss fails against a finite bankroll