Detailed Guide: How the DCA Bot Works

Table of Contents

• 1. Overall Model
• 2. Lifecycle of a Trade
• 3. Buying Process
• 4. In-Depth Analysis of Technical Indicators
• 5. Selling Process
• 6. What Happens When Multiple Mechanisms Are Combined
• 7. Minimum Order Size & Dust Handling
• 8. Supporting Features
• 9. Reliability & State Recovery
• 10. Full Configuration Parameter Table
• 11. Current Limitations to Be Aware Of
• 12. How to Use: Isolate, Backtest, Panic Sell

1. Overall Model

The bot follows a classic DCA (Dollar-Cost Averaging) model: it opens a position with a Base Order (BO), then, if price drops, buys more in stages via Safety Orders (SOs) to lower the average entry price, and exits using one or more take-profit/stop-loss mechanisms running in parallel.

Each trading pair (symbol) is managed by its own independent worker loop, scanning conditions every 0.5 seconds: buy conditions are checked first, then sell conditions, in that exact sequential order on every iteration.

2. Lifecycle of a Trade

[Waiting for buy signal] -> BASE ORDER filled
        |
        v
[Holding a position] ---- price drops ----> SAFETY ORDER 1, 2, ... (if enabled)
        |
        +- price rises enough -> TAKE PROFIT (Static / Dynamic / Trailing)
        |
        +- price drops to threshold -> STOP LOSS (Static / Dynamic)
        |
        v
[Position fully closed] -> state reset -> back to step 1

An important point to grasp before reading the sections below: every buy/sell order in the bot is reactive. The bot does not pre-place any resting orders on the exchange order book. It continuously watches the price, and only once a condition is detected as met does it submit a LIMIT order priced exactly at the market price observed at that moment (deliberately, so it fills immediately, accepting the Taker role). This applies to both buy orders (BO/SO) and LIMIT sell orders used for take profit (Static TP, Dynamic TP).

3. Buying Process

3.1. Base Order: entry conditions

The bot only buys a Base Order when all enabled technical indicators are satisfied at the same time (a logical AND, not OR):

Indicator	Buy condition
RSI	rsi_min < RSI < rsi_oversold (default: 20 < RSI < 30)
MA	Short-term MA crosses above long-term MA (ma_short > ma_long)
MACD	MACD line above the Signal line
ADX	ADX > adx_threshold and +DI > -DI (uptrend strong enough)
Bollinger Bands	Current price touches or is below the lower band
Volume	Trading volume exceeds the configured threshold

Important note: if you do not enable any indicator, the bot will buy the Base Order immediately once the capital condition is met, without waiting for any technical signal. This is deliberate behavior, and the bot will send a one-time alert when it detects this situation.

3.2. How buy orders are placed and tracked

1. The bot places a LIMIT BUY order priced exactly at the market price observed at the moment it decides to buy (not using Post-Only, to avoid the exchange rejecting the order; it accepts filling immediately as a Taker).
2. After placing the order, the bot waits up to 120 seconds, checking fill status every 3 seconds.
3. If the price is still close to the order's price (within 0.2%), the bot extends the wait by another 60 seconds, up to 5 extensions (about 7 minutes total).
4. If the price has moved away from the order's price without filling, the bot cancels the order and, if the buy signal is still valid, retries at a new price for the remaining unfilled amount.
5. If the order is only partially filled before being cancelled, the filled portion is still fully recorded into the position; only the unfilled portion is retried.

3.3. Safety Orders (DCA): lowering the average entry price

When price drops after the Base Order, the bot buys more in configured SO tiers:

• Trigger level for each SO: The parameter uses an array of percentages strictly anchored to your original Base Order price. For example, setting [1.0, 2.0] means: SO1 triggers when the market price drops 1% below your initial Base Order fill price. SO2 triggers when the price drops 2% below that same original Base Order price. By referencing the initial entry rather than the moving average cost, the bot maintains your exact intended spacing and prevents the safety net from compressing as your average cost lowers.
• Size of each SO: The capital deployed for every Safety Order is calculated as allocated_capital × safety_order_volume_scale. This multiplier is applied as a fixed, static value across all SO tiers–it does not compound or scale exponentially (Martingale) with subsequent orders. Each dip-buying order will deploy the exact same amount of USDT. (Note: If you previously used a progressively scaling Martingale strategy, please note this version utilizes a linear volume approach to strictly control maximum capital exposure).
• Required conditions before buying an SO:
  • No Take Profit tier (TP1/TP2/TP3/Dynamic TP) has ever triggered for this position.
  • At least 60 seconds have passed since the last buy.
  • If RSI is enabled (Smart Filter): RSI must be greater than rsi_min (default 20). If the RSI drops below this threshold, the bot will refuse to buy any Safety Order regardless of how far the price has dropped. This is a crucial defense mechanism that prevents the bot from blindly buying into a sharp market crash (deep oversold condition) just because a grid percentage was hit, saving your capital from "catching a falling knife."
  • Sufficient USDT balance is available.
• Once any Take Profit tier has fired (even just TP1), the entire Safety Order mechanism is permanently locked for that position. The bot will never buy another SO again, even if price later drops sharply through the remaining SO levels.

4. In-Depth Analysis of Technical Indicators

This section covers the meaning, strengths/weaknesses, and practical usage of each indicator the bot supports.

4.1. RSI (Relative Strength Index)

• Formula used: A Wilder-style moving average (EMA with alpha = 1/period). Smoother and reacts slightly slower than a "raw" RSI formula.
• Meaning: Measures the speed and magnitude of recent price changes, helping identify overbought/oversold zones.
• Buy condition: rsi_min (20) < RSI < rsi_oversold (30). The bot deliberately avoids buying both when RSI is too low (treated as a risk that price keeps falling) and when RSI has already recovered above 30.
• Strengths: Fast to react, works well in sideways / range-bound markets where price repeatedly bounces.
• Weaknesses: In a strong downtrend, RSI can stay in low territory for a long time ("RSI pinned at the bottom"). Buying on RSI alone here easily leads to catching a falling knife.

4.2. MA (Moving Average) crossover

• Formula used: EWMA (EMA) with short/long periods; buy condition is short MA > long MA.
• Meaning: Filters out short-term price noise, confirms the general direction of the trend.
• Strengths: Works well in a clearly trending market.
• Weaknesses: A lagging indicator; in a sideways market, the two MA lines cross back and forth repeatedly, generating false signals (whipsaw).

4.3. MACD

• Formula used: Standard 12/26/9 (EMA12 minus EMA26 as the MACD line, EMA9 of MACD as the Signal line).
• Meaning: Combines both trend and momentum.
• Buy condition: MACD crosses above Signal, an emerging bullish momentum signal.
• Strengths: Captures both trend and accelerating momentum, reasonably reliable in moderately trending markets.
• Weaknesses: Still a lagging indicator; in a tight sideways range it tends to cross back and forth repeatedly near the zero line.

4.4. ADX + DI (Average Directional Index)

• Buy condition: ADX > adx_threshold (25) and +DI > -DI.
• Key distinction: ADX does not indicate whether the trend is up or down; it only measures the STRENGTH of the current trend. This is why the bot must also combine the +DI > -DI condition to determine the direction of the trend (up).
• Strengths: A very good filter to avoid entering during a weak, sideways market.
• Weaknesses: A fairly heavy lagging indicator; it typically only confirms a trend after it has already run for a while.

4.5. Bollinger Bands

• Buy condition: Closing price touches or falls below the lower band.
• Strengths: Very effective in a sideways market with stable range.
• Weaknesses: In a strong trending downtrend, price can "walk the band." Buying every time price touches the lower band in this situation means repeatedly buying into an accelerating downtrend.

4.6. Volume

• Buy condition: (current volume / 20-candle average) x 100 > volume_threshold (150.0).
• Meaning: Confirms the "genuineness" of a price move.
• Strengths: A very good confirmation filter, helping filter out false signals from RSI/MA/Bollinger when liquidity is weak.
• Weaknesses: Volume alone does not indicate the direction of price; it must be used alongside at least one direction-determining indicator.

4.7. Closed candle vs. open/live candle

The use_closed_candle_only parameter determines whether the bot calculates indicators on fully closed candles or on the candle currently forming (live).

	Closed candles only (default, True)	Allow using the live candle (False)
Pros	Stable signals, no "repainting" of values; RSI/MA/MACD values are final.	Reacts faster, may catch an entry point earlier during a sharp move.
Cons	Always lags real price action by up to one full candle.	"Flickering" signals risk entry based on an unstable state.

5. Selling Process

The bot has 5 exit mechanisms, each can be toggled independently and they run in parallel.

5.1. Static Take Profit: fixed staged profit-taking

Split into 3 levels, TP1, TP2, and TP3, each selling a portion of the position according to configured percentages. The actual sell amount is always capped by the real quantity currently held, ensuring the position is never stuck with leftover coins after all 3 tiers have triggered, as long as tp3_sell_percent is kept high enough (100% is safe).

If a tier's calculated amount comes out below 5.5 USDT (under the exchange's minimum), the bot automatically merges that portion into the next tier.

5.2. Dynamic Take Profit: profit-taking based on ATR volatility

The take-profit level is calculated dynamically: entry price + (ATR x atr_tp_multiplier). If Static TP is also enabled at the same time, Dynamic TP is completely locked out until TP3 (static) has triggered. If Static TP is disabled, Dynamic TP operates independently from the start.

5.3. Trailing Take Profit: profit-taking based on the price peak

Instead of selling at a fixed price, Trailing TP follows the price upward to maximize profits during strong pumps.

Example (Min Profit = 2.0%, Trailing TP = 1.0%):

1. Activate: The bot waits until the price reaches +2.0% profit. It does NOT sell yet.
2. Track Peak: As the price climbs, the exit trigger moves up with it, always staying exactly 1.0% below the highest peak.
3. Exit: When the price peaks and drops exactly 1.0%, the bot instantly executes a MARKET SELL to close the position, letting you ride the full wave.

⚠️ WARNING: MARKET ORDER FEES & SLIPPAGE

Trailing TP uses a MARKET SELL order to ensure immediate execution when the price drops.

1. The Risk: Min Profit does NOT account for exchange trading fees (typically 0.1%-0.2%) and market slippage. If set too low (e.g., 0.1% or 0.2%), your net return could be NEGATIVE (a loss) after fees.
2. Recommendation: Always set Min Profit to >= 0.5% to guarantee a positive net profit after all costs are covered.

5.4. Static Stop Loss & 5.5. Dynamic Stop Loss

Sells the entire position with a MARKET order when price drops to either the fixed percentage (Static) or the ATR-based level (Dynamic). The bot cuts the loss as soon as either condition is hit first. There is no Trailing Stop Loss feature.

5.6. Panic Sell: manual emergency exit

The user presses a button to immediately sell an open position with a MARKET order. If you select a specific pair, it sells only that pair. If no pair is selected, it triggers a 'Panic Sell All' for your entire active portfolio. After a successful Panic Sell, that pair is automatically paused to prevent the bot from immediately buying back in.

5.7. Isolation: removing a position from automatic DCA

Allows you to "isolate" a currently held position out of the DCA management system and set a manual target sell price. The position moves to a separate tracking list, checked every 5 minutes, freeing the bot to resume trading that pair from scratch whenever a new signal appears.

6. What Happens When Multiple Mechanisms Are Combined

6.1. Take Profit triggering permanently locks out Safety Orders

Because every Safety Order is placed reactively, there is no SO order sitting on the exchange to "cancel". Instead, a state flag is set. If any TP tier has ever triggered, the bot will not buy more to lower the average price even if the price reverses sharply. This avoids a conflict between "currently taking profit" and "currently averaging down a loss".

6.2. Static TP + Dynamic TP

Static TP always takes absolute priority. With the default configuration (tp3_sell_percent = 100%), the position is always fully sold at TP3, so in practice Dynamic TP never gets a chance to run. Recommendation: if you choose Static TP, disable Dynamic TP entirely.

6.3. Dynamic TP + Trailing TP

Dynamic TP only sells dynamic_tp_sell_percent% once the ATR-based TP level is reached, and hands the remainder to Trailing TP, which continues tracking the price peak.

6.4. Trailing TP + Stop Loss

Because Trailing TP has the constraint of "never selling at a loss", in a scenario where price rises slightly then drops sharply, Stop Loss (if enabled) is the only line of defense preventing the position from losing further.

6.5. Consecutive losses automatically pause a pair for 24 hours

When a pair reaches 3 consecutive stop-loss exits, the bot automatically pauses trading for that pair for 24 hours from the most recent loss, then automatically resumes.

7. Minimum Order Size & Dust Handling

• Binance requires every buy/sell order to be at least 5 USDT. If a tier's calculated amount is below this, the bot automatically merges that portion into the next tier.
• If the coins remaining after a sale fall below the "dust" threshold (5 USDT), the bot treats the position as fully closed and automatically resets its state.

8. Supporting Features

• Insufficient balance warning: Sent through the GUI if available USDT is not enough to open a Base Order or the next Safety Order.
• Risky configuration warning: The interface automatically checks and warns if: the Base Order capital is too small, any TP tier's amount would fall below the exchange minimum, or the total percentage across the Static TP tiers is below 100% (risk of stuck positions).

9. Reliability & State Recovery

• Each position's state (coins held, average entry price, TP flags triggered) is periodically saved to disk and automatically restored on server restart.
• Market price is fetched preferentially from a real-time websocket, with fallback layers (REST API) if the primary source is interrupted.
• Every API call error has an automatic retry mechanism with increasing backoff.

10. Full Configuration Parameter Table

Parameter	Default	Meaning
capital	100.0	USDT capital allocated to this pair's Base Order
interval	15m	Candle timeframe used for indicator calculations
use_closed_candle_only	True	Only calculate indicators on closed candles (avoids noise mid-candle)
use_rsi / rsi_period / rsi_min / rsi_oversold	True / 14 / 20 / 30	RSI zone condition allowing a buy
use_ma / ma_short_period / ma_long_period	False / 9 / 21	Moving average crossover condition
use_macd	False	MACD crossing above Signal condition
use_adx / adx_period / adx_threshold	False / 14 / 25	Trend strength condition (+DI > -DI is hardcoded)
use_bollinger / bollinger_period / bollinger_std	False / 20 / 2.0	Price touching the lower Bollinger band condition
use_volume / volume_threshold	False / 150.0	Minimum trading volume condition
max_safety_orders	2	Maximum number of Safety Order tiers
safety_order_steps	[1.0, 2.0]	Percentage drop to trigger each SO tier
safety_order_volume_scale	1.0	Multiplier applied to capital to get each SO's size
use_static_tp	True	Enable/disable fixed staged profit-taking
tp1_percent / tp1_sell_percent	2.0 / 40.0	Profit level % and sell % at TP1
tp2_percent / tp2_sell_percent	4.0 / 50.0	Profit level % and sell % at TP2
tp3_percent / tp3_sell_percent	6.0 / 100.0	Profit level % and sell % at TP3
use_dynamic_tp	False	Enable/disable ATR-based profit-taking
atr_tp_multiplier	2.0	ATR multiplier used to calculate the dynamic profit-taking level
dynamic_tp_sell_percent	50.0	Sell % at Dynamic TP (only meaningful when Trailing TP is also enabled)
use_trailing_tp	False	Enable/disable peak-based profit-taking
trailing_take_profit_percent	1.0	Percentage price is allowed to pull back from the peak before selling
min_profit_percent	0.5	Minimum required profit before Trailing TP or Dynamic TP is allowed to sell
use_static_sl / static_sl_percent	False / 5.0	Fixed-percentage stop loss
use_dynamic_sl / stop_loss_atr_multiplier	False / 2.0	ATR-based stop loss
atr_period	14	ATR calculation period (shared by Dynamic TP and Dynamic SL)

11. Current Limitations to Be Aware Of

These are architectural characteristics users should know before trading with significant capital:

1. No SO orders are pre-placed on the exchange order book. The bot detects and places orders reactively (polling every 0.5 seconds), so during extremely fast price moves, it is possible to miss an SO level if price sweeps through it and recovers within under a second.
2. Every buy order and every LIMIT take-profit sell order is priced at the current market price to guarantee an immediate fill, meaning they almost always execute as a Taker, incurring higher fees than resting orders would (Maker).

Prioritizing Taker orders results in slightly higher transaction fees for the user. However, in the architecture of an automated trading bot (especially a DCA grid), the practice of holding orders internally on the server and only executing Taker orders when conditions are met is not a sign of poor programming; rather, it is a deliberate trade-off made to secure the following four critical strategic advantages:

• Guaranteed execution: During periods of extreme market volatility (such as flash crashes or sudden price spikes), the price might touch a Maker Limit level for just a millisecond before reversing. This often results in no fill at all because the order is queued behind others. With this design, once the current price hits the threshold, the bot immediately fires a limit order close to the market price, consuming the opposing liquidity directly and ensuring the position is 100% closed or opened.
• A necessary foundation for dynamic features: Advanced features like Trailing Take Profit or Dynamic SL/TP require the algorithm to continuously recalculate exit price levels every second. This architecture serves as the backbone for dynamic indicators to function.
• API limit optimization: The bot makes only a single API call to read the price (Stream) and a single POST request to place a buy or sell order.
• Capital and assets remain free: Funds are not locked up on the exchange in pending orders.

3. Once any Take Profit tier has triggered, Safety Orders are permanently locked out for that position, so the remaining position after a partial profit-take is not any better protected against subsequent adverse moves, beyond whatever Stop Loss you may have enabled.

12. How to Use: Isolate, Backtest, Panic Sell

12.1. Panic Sell: manual emergency exit

When to use it: you want to exit one or all open positions immediately, regardless of what the indicators/TP/SL are currently saying.

Steps:

1. Select the pair you want to exit from the list on the left. If no pair is selected, it triggers a 'Panic Sell All' for your entire active portfolio.
2. Click the PANIC SELL button.
3. Confirm the warning dialog that appears.
4. The bot immediately sends a MARKET SELL order for the entire quantity held of that pair.

Things to note after clicking it:

• Panic Sell only sells the asset amount the bot itself is tracking. It does not scan and sell your entire wallet balance on the exchange.
• After a successful sale, that pair automatically switches to a paused state (is_running = False) to prevent the bot from immediately buying back in.
• Because it uses a MARKET order, the actual fill price may slip relative to the price displayed.

12.2. Isolate: removing a position from automatic DCA management

When to use it: you want to "shelve" a position that is losing money, freeing the bot to resume trading that pair from scratch whenever a new signal appears.

Steps to isolate a position:

1. On the real-time status table, right-click the row of the pair you want to isolate.
2. Select "Isolate this pair" from the menu that appears.
3. Confirm the dialog. Note the warning: this pair will stop DCA entirely.
4. Enter the target sell price (USDT). By default the bot suggests a starting reference of current market price x 1.05.
5. If you enter a target price lower than or equal to the current market price, the bot will show a separate warning: doing so will cause the position to be sold immediately with a MARKET order.

12.3. Backtest: simulating a strategy on historical data

When to use it: before turning on a new configuration with real money, you want to estimate how it would have performed based on actual historical price data.

Steps:

1. Select the pair you want to test from the list on the left.
2. Click the BACKTEST button.
3. In the dialog that appears, choose the time period and the simulated capital (USDT).
4. Click "LOAD DATA & START BACKTEST".
5. The results appear as a detailed log per trade, ending with a summary: total net profit (after an assumed round-trip fee of 0.2% per buy-sell cycle), number of complete buy-sell cycles, win rate, and total number of filled orders (including both Base Order and Safety Orders).

Key points to understand before trusting the backtest results:

• The backtest uses exactly the configuration currently shown on the form at the moment you click the button, even if you have NOT yet saved that configuration to the server.
• The backtest is a simplified simulation and does not fully reflect every behavior of the real running bot. It calculates indicators over the entire available historical dataset (no data latency), assumes a flat 0.2% round-trip fee per cycle, and uses an approximate dust threshold.
• The backtest does not simulate the latency/risk of missing an order caused by the reactive order-placement mechanism. In real live trading, results will typically be somewhat worse than the backtest figures.
• For these reasons, backtest results are best treated as a relative reference for comparing different configurations, rather than as a reliable absolute profit figure.

This document will be updated whenever the bot's architecture changes.