Omada

4.99

Alpha and Sharpe Ratios according to Morningstar:

Ticker; 3 Year, 10 Year Alpha; 3 Year, 10 Year Sharpe Ratio

PBP 1.84, -2.44; 1.23, 0.5

VEGA -1.78, -3.5; 0.95, 0.51

XYLD 1.58, -2.04; 1.19, 0.57

QYLD 2.7, -0.56; 1.26, 0.69

FTHI 0.13, -2.7; 1.14, 0.54

FTQI 0.95, -1.76; 1.22, 0.56

In other words, all but VEGA have done well in the past 3 years but have done quite poorly over the past 10.

Someone out there thought that was a good picture of her to post on the internet. Yeesh

That's cute, SEC. But before you do anything else, just remember we're ready to spit on our hands, hoist the black flag, and begin slitting throats.

Taken from dailyspeculations.com; posted by Steve Ellison in 2024 and called the wall of worry.
EDIT: I'm posting the chart without my own commentary.

The vast majority of my backtesting has been with 1D or larger data, not intraday. In my opinion, swing trading edges are pretty easy to find compared to intraday ones. Maybe one day I'll get back around to doing intraday backtesting just to see what I can find, but it's something I don't need to do at this point.

quote:

---

On my own journey I have tested pretty much every major statistical catagory, research paper theory, or social media trader setup over the full databento historical CME and OPRA datasets to no avail (29 of 29 kills on backtested theories

---

Are you saying you've only done 29 backtests?

I'm not trying to be mean, but you know you just posted what is basically chat room advertising? The poster has a link to a $75 a month chat room in his bio that, based on current subscribers, makes him and his business partner $18k a month. The vast majority of the reviews are from Indian accounts, so they're likely bought or made by them. And supposedly on that site (whop.com), you can take down someone's review and then remove their membership so they can't repost a negative review, so... yeah. That could explain how virtually no bad reviews exist.

That's not even mentioning his pinned tweet is of a portfolio overview with hidden money amounts. He's up 113%, but you've no idea if he did so with $100 or $100k. Well, you somewhat do because he'd definitely show off doubling the $100k.

quote:

---

You would think WSJ could do a little research.

---

If they devoted a tenth of their subscriber retention efforts to article research, the WSJ would be the most factual news service ever known to man. Instead, it's a paper supposedly for businessmen and members of Wall Street that describes the basics of options every time they're mentioned in an article as if the readers are completely ignorant.

According to an email I received several hours ago, myfxbook.com now offers free backtests for, well, forex. I'm uncertain of their data quality, but if it is good data, then that is a forex option for you since I didn't have one when I made my original post.

EDIT: never mind, it looks like it is just a manual backtest. The site does have a "Strategy AI" that can create strategies for you and will backtest the results, but that is a premium option. I don't think either is worth it, personally. Apologies for bumping the thread.

For day trading "with a humble few thousand," you may be better off instead swing trading. The reason for this is that you need a definite, known edge or else you've no idea if what you're doing has a long-term positive expectancy (that is, it works). To have a definite, known edge, you'll need to backtest, and intraday data to do so can quickly get expensive (it can potentially be expensive for daily data, for that matter), so that can end up eating a big chunk of your few thousand before you begin or require you to put more up.

You'll need to decide what you'll be trading, what rule(s) will affect your chosen security/ies, and obtain at least a basic understanding of said secruity/ies and their market(s). For example, the wash sale rule, which LChama rightly brought up, applies to stocks but not to Section 1256 securities.

As you're trying to decide what security/ies to trade, remember that you'll need historical data for backtests that is not too expensive while also being fairly accurate. Databento is typically inexpensive and high quality, but only goes back so far (May 2018 for equities) and doesn't have forex data. They give $125 in free credits (good for 6 months) to new accounts that you can use on the usage based option. Equity data per symbol can be as low as 1 cent (1D data). For the intraday data, you'll want to select the zstd compressed option during your request, at least if you use CSV format. If you decide to swing trade equities, you may be able to just scrape the 1 day, 1 week, and/or 1 month data from Yahoo Finance or investing.com, though you'll want to make sure the dates are accurate (I've found them off by a day in the past). I've no idea where to get forex or crypto data; investing.com has forex data IIRC, but I couldn't confirm its accuracy.

As for tools to do the backtests, I do mine in Excel. Some trading platforms, websites, and/or programs are available to do the backtesting and may come with data already. However, these may require coding abilities and/or payment to use. An LLM can write code for you, but you will always want to make sure the code works as you want it to. Remember, this is your money on the line.

As for ideas to backtest, the possibilities are endless. You can try common technical indicators and setups (prepare to be disappointed most of the time). You can browse SSRN's eLibrary under the economics research, financial & investment planning research, and financial economics sections under social sciences for ideas others have written papers on. I'll mention that, after a trading method or edge has been used for 10 or 15 years in the industry, someone will often write a paper on it, though that does not apply to most papers, and it doesn't mean that you'll have the resources to use it. You could just ask questions like "what happens after security XYZ declines by 3% or more in one day?" The trick is making sure your question can lead to a quantified result so you know what you're working with.

Once you have the results to your question, you'll need to compare them to a benchmark such as, but not limited to, the S&P 500 or whatever is relevant to your chosen security/ies. You'll want to calculate things such as win rate, trade length, mean and median returns, average and max gains and losses, profit factor, maybe payoff ratio, max drawdown, ulcer index, possibly Sharpe Ratio, and Jensen's alpha (if applicable). LLM's are often your friend here, but remember they make mistakes. Keep in mind that just because your total return is smaller than your benchmark does not necessarily mean it's a bad trade: you may be doing better in terms of time in the market and/or risk exposure, and so some leverage may boost your results to make it worthwhile. You'll need a rule or rules to stop losing trades from running too long. You'll also need to deduct from your backtested results to account for trading fees and slippage (where you get in and out vs where you are trying to). If you do find something, you can then try paper trading it and then live trading.

Or, after reading all this, you could just say F it and use the money to bet on horses for fun instead.

EDIT: I forgot you'll want to calculate your idea's and your benchmark's CAGRs for the sample period, and calculating CDGR (daily instead of annual) is great to see how well a system is doing while active. For ideas, the paper (later made a book) 151 Trading Strategies provides a number of ideas, some of which you will find either not easily testable or not significantly worthwhile. I changed 1 (not saying which) and gave the SPX backtest results to a friend for her to potentially use. With my changes, the 2x leveraged CAGR was about 14% over the past 30 years, so I know for a fact at least 1 idea with some tweaks is worthwhile. And whatever you do, DON'T OVERFIT TO YOUR DATA!!!

You may want to read Giuseppe Paleologo's Advanced Portfolio Management: A Quant's Guide for Fundamental Investors. I want to be clear that it is not a light read, but based on what you've said in your post, you may be able to handle it. It may not provide a definitive answer to each of your questions because the solutions may not be conclusive or simple, but it will expose you to some options and discuss them.

One example I'll mention is position sizing. Using mean variance for position sizing, aka constructing a Markowitz portfolio (modern portfolio theory), often leads to "highly unintuitive allocations" and/or underperformance. His recommendations include a proportional rule (more alpha = more allocation), a 1/N rule, or a mean variance with a 75% shrunken asset variance. Someone backtested the methods with crypto here if you are interested or want to read a bit more about them. 1/N is simple and does well enough if the other rules seem like too much work.

It's not a perfect book, but it certainly beats an LLM potentially telling you to invest using half Kelly even though that is often too aggressive and doesn't factor in max drawdown.