The Quest Begins (The "Why")
I still remember the first time I fired up a tiny crypto‑trading bot in my bedroom. I’d just finished a tutorial that promised “make money while you sleep,” slapped together a few API calls, and hit run. The console lit up with green “BUY” signals, and I felt like I’d just discovered the One Ring.
Five minutes later the market moved against me, my position was half‑filled, and the bot kept screaming “BUY” even though I already owned the asset. My P&L swung from a tidy profit to a painful loss faster than a podrace on Tatooine. I stared at the screen, wondering where the force had gone.
That moment kicked off a quest: what are the sneaky traps that turn a promising trading system into a fragile glass sword? I dug into logs, read exchange docs, and (after a few too many 3 a.m. debugging sessions) uncovered a handful of mistakes that keep tripping developers up—even seasoned ones. Let’s arm ourselves with the knowledge to avoid them.
The Revelation (The Insight)
The biggest eye‑opener was realizing that most failures aren’t about exotic math or secret indicators. They’re about the mundane assumptions we make when talking to an exchange:
- Assuming an order fills instantly and completely.
- Sending market orders without checking slippage or the order‑book depth.
Both seem harmless until volatility spikes or the market is thin. The bot then ends up buying at a price far worse than expected, or it keeps re‑sending orders because it thinks the previous one never filled.
Once I stopped treating the exchange like a magic vending machine and started respecting its real‑world quirks, my bot’s behavior became predictable, survivable, and—dare I say—profitable.
Let’s look at the code that caused the trouble, and then the spell that fixed it.
Wielding the Power (Code & Examples)
Trap #1: “My order filled, right?”
Before – the fragile assumption
import ccxt
import time
exchange = ccxt.binance({'enableRateLimit': True})
def place_market_buy(symbol, amount_usd):
# Grab the current price and convert USD to base currency
ticker = exchange.fetch_ticker(symbol)
price = ticker['last']
amount = amount_usd / price # <-- assumes we can buy this amount
order = exchange.create_market_buy_order(symbol, amount)
print(f"Market BUY placed: {order['id']}")
# …later, we assume the whole amount is ours
position = amount # WRONG! Only part may have filled
return position
If the market is moving fast or the order book is shallow, create_market_buy_order might only fill a fraction of amount. The bot then believes it owns the full quantity, skews its risk calculations, and may double‑buy on the next loop.
After – respecting partial fills
def place_market_buy_safe(symbol, amount_usd):
ticker = exchange.fetch_ticker(symbol)
price = ticker['last']
amount = amount_usd / price
order = exchange.create_market_buy_order(symbol, amount)
order_id = order['id']
# Poll the exchange until the order is closed (filled or cancelled)
while True:
order_status = exchange.fetch_order(order_id, symbol)
filled = order_status['filled'] # base currency actually received
remaining = order_status['remaining']
if order_status['status'] in ('closed', 'canceled'):
break
time.sleep(0.2) # avoid hammering the API
print(f"Order {order_id} filled {filled}/{amount} ({filled/amount*100:.1f}%)")
return filled # <-- real amount we now hold
Now we query the order status until we know exactly how much was filled. The rest of our logic uses filled, not the original amount. This tiny change stops the bot from hallucinating phantom positions and keeps risk metrics honest.
Trap #2: “Market orders are free!”
Before – slippage blind‑fold
def aggressive_market_sell(symbol, amount):
# Just hit the market, no regard for depth
order = exchange.create_market_sell_order(symbol, amount)
print(f"Sold {amount} at market price")
return order
When the book is thin, a market sell can walk through several price levels, selling a chunk at a far lower bid than the mid‑price you saw moments ago. The resulting slippage can turn a “small profit” trade into a loss.
After – checking depth and using a limit order with a tolerance
def safe_sell_with_slippage_limit(symbol, amount, max_slippage_pct=0.2):
"""
Place a limit order that will only execute if the worst price
is within `max_slippage_pct` of the current mid price.
"""
order_book = exchange.fetch_order_book(symbol)
best_bid = order_book['bids'][0][0] if order_book['bids'] else None
best_ask = order_book['asks'][0][0] if order_book['asks'] else None
if not best_bid or not best_ask:
raise RuntimeError("Order book empty!")
mid_price = (best_bid + best_ask) / 2
worst_acceptable_price = mid_price * (1 - max_slippage_pct / 100)
# Ensure our limit price is not worse than we tolerate
limit_price = min(best_bid, worst_acceptable_price) # sell at or above this price
order = exchange.create_limit_sell_order(symbol, amount, limit_price)
print(f"Placed limit SELL @ {limit_price:.2f} (mid {mid_price:.2f}, slippage limit {max_slippage_pct}%)")
return order
We look at the order book, compute a reasonable mid price, and then set a limit price that guarantees we won’t sell worse than a user‑defined slippage threshold. If the market moves against us and the limit can’t be filled, the order simply sits open—giving us a chance to reconsider or cancel instead of swallowing a nasty surprise.
Why This New Power Matters
By swapping naïve assumptions for real‑time checks:
- Risk stays predictable. Your position size always reflects what you actually own, not what you hoped you owned.
- Slippage is under control. You define how much price deterioration you’ll tolerate, and the bot respects it.
- The system becomes resilient. In volatile or thin markets, it won’t blow up because it’s waiting for a fill or refusing to execute a dangerous market order.
These patterns aren’t just academic; they’re the difference between a bot that survives a flash crash and one that gets liquidated before you can hit “Ctrl‑C”.
Now you have two concrete “spells” to add to your trading‑system grimoire:
- Validate fills before trusting an order’s quantity.
- Guard against slippage by checking the book and using limit orders with a tolerance.
Apply them, watch your error logs shrink, and feel that quiet satisfaction when your bot calmly weaves through market turbulence like a Jedi deflecting blaster bolts.
Your Turn – The Challenge
Pick one of the traps above (or another you’ve battled) and rewrite a small piece of your own trading code to handle it. Share your before/after snippets in the comments, or tweet them with #TradingSystemQuest.
What’s the first mistake you’ll conquer on your own odyssey? I can’t wait to see the clever solutions you all come up with! Happy coding, and may your fills be full and your slippage low. 🚀
