That's before you factor in taking out some for taxes every quarter and some to pay yourself. My wife got mad at me for "unrealistic exponential math", so I made a model that included taxes, expenses, and withdrawing some out for pay.


February 8 will have to do.



The $ amount on it is misleading, because those bots are trading against BTC. Initial investment was 0.3 BTC. The profit on it is 0.0295 BTC, so a little less than 10% - about 13.2% annualized—measured in BTC. So that's 13% alpha against BTC.

I have other more recent stuff doing better than that — that's just the only thing I have to show running that long. And since it's doing so well on the risk management side (pretty smooth curve), I've kept it running.

This is the common criticism of skeptics.

But let's say you could make 3% a month—definitely profitable trading. Then you need to set aside, say, 1/3 of that for taxes, so it's really only 2% a month. And then, you should pay yourself. Take some of the profits and improve your lifestyle. Maybe also invest in some tools, education, memberships, subscriptions, etc. Let's allocate 20% for that. What's left, let's figure compounding it monthly.

And then let's start with $10,000, even—not where most crypto traders are starting, for sure.

You know how long it takes you to accumulate $1M?

27 years

That's how long it takes to become a millionaire, starting with $10K, attainable profits (that are still way better than the stock market), taxes, expenses, etc.

Now, let's start with $1,000 instead, only 25% for taxes, and don't pay yourself—roll everything back in, and compound weekly.

It's still 25.5 years.

So no, most crypto traders wouldn't be millionaires by now, even if they're consistently profitable. People are only going to become millionaires by:
1. Starting with a lot of money (like $100K),
2. Getting way higher returns,
3. Trading on leverage (crazy risky), and/or
4. Trading with other people's money (prop trading or being a lead copy trader)

Not that hard to just be profitable with crypto bots. Making a million? Yeah, very hard.

No, I understand exactly what it is, and have been doing it, the entire time I've been trading. I just didn't wait for that to start trading live accounts. But I started with just two strategies, with $500 each. Now I'm up to trading my entire crypto portfolio allocation live. 80% goes into strategies that have been proven out by forward testing, whether live or paper trading, and 20% goes into live testing new strategies from backtesting.

There's a difference between preparing for the bad situations and expecting them. Having smoke alarms and a fire extinguisher in your house is preparing for the bad situation.

Good risk management is a must. That's why most traders fail—going all in on something without good risk management, then being unwilling to cut their losses.

And good risk management doesn't mean you miss something good. I caught the DOGE run last week. About 30% on one bot, 40% on another, and one that hasn't exited yet. Now, that's not the 100% that someone could've made perfectly timing an entry and exit, and I only have about 4% of my portfolio in DOGE. But the point is, it all happened automatically. I wasn't watching the news closely and even then, I didn't make the connection that the finalization of the Twitter deal was going to make DOGE surge. I literally did nothing, other than decide to include DOGE in my portfolio a couple of months ago, when I decided it had finally moved out of the realm of $#!+coin.

Risk management isn't fear.

If you fail to plan, you plan to fail.

That hasn't been the case so far.

Forward testing is important. But how long do you run forward testing? And most importantly, what do you do in the meantime? Really, forward testing is just more out-of-sample data. At the end of your forward testing, it's just backtesting, and no better at actually predicting the future from there forward than the backtesting.

The two essential things I found in backtesting are:

1. Test the strategy against out-of-sample data, not just 30/90/180/365 back from today. How did the strategy do from November 2021 through June 2022? How did it do against the market (what's the alpha) from mid-June through mid-August? (BTW, my live strategy did over 60% during that 60-day window, vs. BTC, which went up 29% during that same time period).

2. Don't over-optimize. If you find, for example, that a 9% stop loss is your optimum profitability, but an 8% or 10% stop loss actually puts you at a loss, then that's a bad strategy. Instead, find a range of stop losses that still keep you in profit, and put it in the middle of that range. True example.


About a year, but I have a background in statistics, finance, and modeling. And I'm obsessive about both learning and experimenting. I've been consistently beating the market since the start (not counting a few failed experiments with live data, which I've also learned not to do).

I really do appreciate you sharing your experience, and maybe some day it will ring in my head, "You know, that guy was right." At this point, I'm certainly not ignoring it—just taking it with a grain of salt.

Let me ask you this: you say do forward testing. How long? And what do you trade live in the meantime?

I'm surprised at all this hate for bots. I have a day job, so I can't even imagine trading without them. I certainly don't want to make trading a full-time job, especially not if I can get 80% of the results with 20% of the effort. Or even 50% of the results with 10% of the effort. And I'm not convinced it's even that.

Bots trade 24/7/365, just like crypto. They never miss a signal because it's the middle of the night, or you're at lunch, or on vacation, or whatever. Bots follow all the practices traders have to learn the hard way:
- Don't trade on emotion
- Don't get greedy - take your profits
- No FOMO - you're not going to catch every surge
- Diversification. You can't manually trade 20 coins. Sorry.

Also, this idea that beginners can't do it? There are levels, just like with manual trading.

Here's a simple, simple way that you can beat hodling, and it takes 5 minutes to set up. And you can do it for free on Binance, Pionex, or use 3Commas on any exchange they support. I'm sure there are others.

Set up a grid bot trading or a rebalancing bot between ETH/BTC. You'll accumulate BTC ahead of the market, and you hardly ever have to adjust it, because ETH and BTC are pretty well correlated at the macro level (but not at the micro level). This is what mine looks like. This is only 60 days or so, because I had to liquidate it for something else, but I've been running this basic bot setup for about a year. Beating the market by ~30%.



Other grid bots do even better, especially against BTC. SUSHI, REN, and BAL are all running 5%/month or higher grid trading against BTC.

As far as indicator-driven bots, I've been running those profitably all year too. Sure, I've had some experiments that are losses. My current algorithm backtests on BTC/USD at over a 30% annualized return YTD. Yes, in a bear market. Did even better in that slight uptrend from mid-June to mid-August. Other major coins (not counting ones that had black swan events) have done great this year too, with returns above 5% monthly, some above 15% monthly.

Yes, you still have to have an eye on the coins at a macro level. Still have to keep an eye on the news. But you don't have to be a genius to trade profitably with bots. Just have to apply the same principles -- take your profits, don't get greedy, manage your risk, diversify as your account grows.

BTW, as far as what it costs to get into all this?

- Binance and Pionex have free grid trading - I'm sure some other platforms do too.
- 3Commas free account allows you one multi-pair DCA bot and one grid bot. A multi-pair DCA bot is amazingly powerful. Mine did 30%/month from mid-June to mid-August. Yes, it took some losses after that, but I've learned and tuned it and it's now running profitably, 10 deals at a time.
- CoinRule's free plan lets you run 2 live rules, up to $3K monthly trade volume.
- Trality's free plan allows you one rented bot and one created bot, up to €5,000 trading volume. Plus, unlimited backtesting (great feature).
- CryptoHero and TradeSanta both have free plans with 1 bot.

And, of course, just about every platform has a free trial.

Personally, I use Pionex for my grids, 3Commas free for 1 grid and 1 multi-pair DCA bot on my main exchange account, and CryptoHero for my algorithmic bots. CryptoHero doesn't have as many strategy options as some of the more mature bot platforms, but it's a lot less expensive, has great support, and very fast backtesting (much faster than other platforms).

I suggest just getting started with the free plans on a couple of different platforms. Get something profitable running. Then optimize it. Then go try something different on another platform -- each platform tends to have one or two main approaches that they're best at. It's not an either/or — just because you start experimenting with bots doesn't mean you can't keep trading manually. But at least try it out and give it a fair shake.

I'll leave you with this — screenshot of my main bot account's last few days of trading:

That's what stop loss is for.


Actually, the swing trading strategy I'm using works pretty well in just about any market. I've tested it with out-of-sample data, e.g., the long bear run from Nov 21 to June 22, and it's still profitable. I do selectively turn bots on and off depending on long-term outlooks for the coin, but the goal is to pretty much let them run and trust the statistics.

I like the idea of a weighted split test with live going forward. I always forward test the top few strategies from the backtesting, but yeah, a weighted split test live is a good idea for balancing risk.

And yes, I figured out quickly how essential it is to include fees in the backtesting calculations. Of course, that's another reason to favor the higher frequency strategy: it increases your monthly volume and may help you move to a lower fee tier. Currently my monthly trading volume is about 20x my average daily balance, which takes 0.06% off both maker and taker fees.

Oh, I absolutely forward test too. But backtesting helps you figure out what to forward test, and what to run live in the meantime. You need both.

As a newbie, start with BTC/USD or BTC/USDT. Get to the point you can consistently make money with that. Pretty much everything else is too volatile for beginners.

Bitcoin is now less volatile than the stock market (S&P 500). Its largest single day drop in the current bear market was 16%, back in June. The last time it had a drop that big was March 2020, when Covid hit the US. Its current 30-day daily volatility index is 3.5%. Plenty for making profits, but you won't get rekt in a day or two.

As far as beginning strategies, you might try a swing trading strategy with RSI and Stochastic RSI, 15 minute trading period. with about a 4% stop loss. That's enough to  allow usual fluctuations, but pull you out of big slides.

Depending exactly how you implement it, that strategy over the last 6 months would've done about 36 trades, 75% W/L ratio, total profit of about 12% (24% annually). That's a good starting point to work from. With tuning, you can easily get that up to 30% annually. Just trading BTC, two indicators, fully automated.

Get a feel for the charts first, but then you can automate that on the free plan of 3Commas, CoinRule, CryptoHero, probably some others.

Good luck!

I'm to the point that I'm consistently profitable, now trying to optimize strategies and allocate risk appropriately between different strategies. I'm familiar with basic risk reward ratios. What I'm trying to do is incorporate average deal frequency and duration into my risk management.

Let's say I do 180 day backtest on two strategies. Once fees are figured in, they both produce the exact same profit over the period. But one does it in two deals, the other does it in 20. It seems obvious to me that the one that did it in 20 is less risky, because it's more consistently predictable.

Similarly, let's say two strategies both have 20 deals, but one has average deal duration of 2 hours, the other 2 days. Again it seems obvious that the one with average deal time of 2 hours is better, all other things being equal. There less time for things to go wrong.

But the question is how much to factor that into expected value / risk-weighted return. How much higher return does the 2-deal strategy need to make to be "equal" to the 20-deal strategy?

Is there any standard way for incorporating these into risk assessment? I've looked but haven't found anything.

Or what ideas do you have about factoring these variables in your risk assessment?