Skip to main content

Disclaimer: The information presented does not constitute financial, investment, trading, or other types of advice and is solely the writer’s opinion.

On 14 March 2023, OpenAI announced the release of their newest, smartest bot so far. ChatGPT 4.0 is better than version 3.5 in a multitude of ways. However, even the 3.5 bot can be used to our advantage. What can the bot contribute to a trader whose focus is technical analysis?

The bot proved useful in explaining and presenting information on a variety of tools and price action concepts. When prompted, it can explain the calculations and the usage and interpretation of many of the common indicators that traders use, ranging from RSI to Bollinger bands. It can also provide users with guides on how to begin to analyze the market using the principles of Elliot Wave Theory. While it can not coach advanced concepts by itself, it can be a useful assistant.

One major limitation of ChatGPT is its lack of access to live data. This means that the bot cannot dissect current information such as live prices or other relevant data points of the market. It cannot be used as an alert bot for topics beginning to gain traction quickly, nor can it respond to a considerable flurry of buying volume on a low-cap coin in the crypto-market. However, it can be used to string together TradingView indicators for a beginner trader to come up with strategies.

Taking the help of ChatGPT in devising a basic intra-day strategy

One can come up with an endless array of strategies to trade on various timeframes using a combination of TradingView indicators. The only limitation is the user’s imagination and familiarity with indicators. It is unlikely that ChatGPT can come up with predictions based on data for the prices of an asset such as Cardano. The AI model developed by AMBCrypto, on the other hand, can.


Read Cardano’s [ADA] Price Prediction 2023-24


I began with a fairly simple task for GPT – Take the RSI and the moving averages and use them together to generate buy and sell signals for intra-day traders. After a few trial requests, the scope was narrowed down. Buy only when RSI is above 50, and use the Fibonacci numbers 13 and 21 as moving averages periods. Here was the response the bot presented –

GPT 3.5 and trading performance

Source: OpenAI

And the PineScript code for the same.

GPT 3.5 and trading performance

Source: OpenAI

I tested the strategy ChatGPT came up with on the Cardano chart. Since the point was to use the bot’s help to generate scalp trade signals, the 2-minute timeframe was used. Here are the results –

GPT 3.5 and trading performance

Source: ADA/USDT on TradingView

Long positions were taken only with the RSI above 50, and shorts only with RSI below 50. After an entry into a position, if the RSI crossed the neutral 50 in the direction opposite to the one during entry, the trade was closed since it showed the momentum had shifted.


Is your portfolio green? Check the Cardano Profit Calculator


Many places gave indecisive entries, but the more favorable ones were highlighted in the chart above. The actual realized R: R of the trade is shown in cyan.

In ten hours, this strategy generated signals which theoretically could have yielded +8.98R. However, these results are cherry-picked. In the chart above there were many other entries that the system suggested are shown, especially in those hours when the prices were trading sideways. Multiple trade entries here would have exited at break even or at a slight loss, with trading fees beginning to pile up too.

What separates a good trader from a bad one?

It is possible to go on and on taking different indicators together, altering and tweaking their input values and backtesting their signals. However, we shall move in the direction of risk management. Risk management is what separates a trader from a gambler. It also helps undercut the emotion a trader might feel during a trade. Fear almost always arises when the trader has risked more than they can stomach. This can negatively impact profitability.

Back-testing aside, any profitable trader must be able to limit their losses. Each trader is probabilistically bound to run into a streak of losing trades. Some key elements of risk management ChatGPT identified were diversification, position sizing, stop-loss orders, risk-reward ratio, and risk tolerance. Diversification is necessary because crypto is a highly volatile market. The assets are, for the most part, positively correlated with Bitcoin. This means that investors could look to allocate only a minority of their funds toward crypto-assets, which would be anywhere from 5% to 50%. Having one’s net worth in crypto is highly risky.

Stop-loss orders are orders placed at levels of invalidation of a trade idea. They are automatically executed and are set up in such a way that the trader exits their losing position if the price reaches a predetermined level. This level can be determined by technical analysis. The capital lost during that trade would ideally be less than 3% of the entire account size. But why? Why shouldn’t one trade by risking a significant chunk of their account size in each trade?

A bad streak in the markets shouldn’t destroy your trading account

GPT 3.5 and trading performance

Source: NewTraderU

The attached chart reveals that a trader with a 30% win rate is guaranteed (has a 100% chance) of having a losing streak of 8 trades within a 100-trade sequence. If the trader risked 10% of their starting account size with each trade and lost eight in a row they would be down by 80%. The trading system isn’t broken, but probability will spoil your profits. Trading is not a sprint to the finish line but an excruciating marathon where your biggest enemy is yourself – Fear and greed, in particular.

To survive, the amount of capital risked per trade must be able to withstand a losing streak, which will be based on the win rate. Even if the trades you take are amazing with 3:1 or 4:1 risk-to-reward, it doesn’t do a lot of good in protecting your capital when the market seemingly has your number.

Hence, risking no more than 1%, or 3% per trade would be far more likely to succeed in the long run. The profits might not be quick, but they will be present. And, the emotional side of trading will also likely lose its intensity since each trade won’t make or break you.

Understanding R: R and calculating when a trader is at break-even

Let’s assume we have an account worth $1000. We are determined to lose no more than 1% per trade, which means each losing trade will only cost $10 or 1% of the total size. Meanwhile, our winning trades could make $20 or $30, or any other amount. The ratio of the capital risked to the reward gained if the trade ran to completion is called risk-to-reward, or R: R. Usually, traders target a 3:1 ratio, meaning they are willing to lose 1% per trade but seek to win 3% of their account size.

A trader will likely not be successful 100% of the time. If they are correct about 30% of the time, they could still be profitable. Even a trader with a 5%-win rate could be profitable in the long run. A trader who only places 3:1 RR trades will need to be successful (1-(3/(3+1))*100 i.e. 25% of the time only to break even. Similarly, a trader who only wins 5% of the time would need to place only trades with an RR of 20:1. (1-(x/x+1))*100=5, solving for x, we get 20.

If a trader looks for 3R trades and has a sound reason (Based on technical analysis or fundamental analysis, for example) to place that trade, and they are successful with more than 25% of their trades, then they will be a profitable trader.

It can help to maintain a Trade Journal

Pesky algebra aside, how does a trader track their win rate? The most common solution is a trading journal. This is a ledger where a trader can jot down each trade they place and the insights they learn from it. ChatGPT can help create a basic template for this purpose –

GPT 3.5 and trading performance

Source: OpenAI

In this template we see the R: R of the trades taken, their success rates, and the trader’s reasons to enter and exit the trade. Traders can also note their emotions to prevent the same mistakes from repeating. The journal can also be used to find an edge in the market for yourself.

This means information about what kind of trade works most often for you. Long or short? If long, could the ones where RSI>50 on both M5 and M15 be another factor you want to check before looking to take long positions since this confluence appears to give your trades more success?

Calculating the capital risked per trade

These and more questions can be answered by implementing a journal. Another tool that the ChatGPT bot can aid in creating is a position size calculator. We have already seen R: R and the success rate can be determined through journals. Let’s try to recall the probability chart presented earlier. Even with a 60% win rate, there is still a 92% probability that one will see a streak of 4 consecutive losing trades within a span of 100 trades.

Therefore, the requirement would be to risk 1% or 3% or something in between for each trade setup. Calculating this can be time-consuming. I asked ChatGPT to come up with some code to help calculate the position size. It obliged and presented below is some code the bot generated. The input prompts would have to be account size, leverage used, risk threshold, and stop loss distance.

GPT 3.5 and trading performance

Source: OpenAI

Let’s assume an account size is $1000, the risk threshold is 5%, stop-loss percentage distance of trade is 6%. The leverage used is 10x. The initial margin required is calculated as:

Margin = (1000 * 0.05) / (0.08 * 10) = $62.5.

For spot traders, the leverage utilized would be 1x.

Exactly how useful is ChatGPT to professional traders?

I asked Mikaela Pisani, ML Lead and Senior Data Scientist at Rootstrap. She is an expert in big data development and artificial intelligence and her response was,

“Traders can use ChatGPT as a tool to get recommendations on the stock market. It is likely to be most useful for beginner traders, enabling them to learn the fundamentals of stock trading from the chatbot. More advanced traders can use it as a tool for gathering insights and making decisions faster, but there are limitations given the output is based on data provided (currently training data is up to 2021).”

As highlighted earlier in the article, the use of the bot in live trading is severely curtailed. But what about the bot’s impact on algorithmic trading?

“Aside from data limitations, which are the primary weakness of ChatGPT for traders, the advantage for traders will be an extremely short window of time as the market absorbs these AI tools to improve efficiency of the market via automation and improved outputs of trading algorithms.

In this way, we can view ChatGPT as likely to have a similar impact to the first High Frequency Trading platforms – yielding a potential advantage for early traders but quickly becoming part of the norm of the market.”

Therefore, a trader will likely not be able to build a lasting edge in the market using ChatGPT alone. However, it can be fantastic in providing tools to traders just beginning their journey. Newer traders will likely be able to use the bot to navigate trading tools much quicker than before. An understanding of basic risk management is also likely to help.




Source link

Leave your vote

Blockchain Beat

Author Blockchain Beat

More posts by Blockchain Beat

Leave a Reply

Your email address will not be published. Required fields are marked *

GIPHY App Key not set. Please check settings

close

Log In

Forgot password?

Don't have an account? Register

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.