{"id":175,"date":"2016-10-09T01:51:59","date_gmt":"2016-10-09T07:51:59","guid":{"rendered":"http:\/\/www.blog.jay-greco.com\/wp\/?p=175"},"modified":"2016-10-09T16:31:06","modified_gmt":"2016-10-09T22:31:06","slug":"analyzing-the-performance-of-acorns-investment-portfolios-using-quantopian","status":"publish","type":"post","link":"http:\/\/www.blog.jay-greco.com\/wp\/?p=175","title":{"rendered":"Analyzing the Performance of Acorns Investment Portfolios using Quantopian"},"content":{"rendered":"

I’ve been using Acorns<\/a>, the app to “help anyone invest” for a little over a year, mainly as a curiosity. One common theme I’ve seen amongst Acorns users is the disappointment with the lack of returns throughout their usage. Simply searching “Acorns Returns” online brings up hundreds of posts (many of them,\u00a0uncoincidentally, occurring on r\/investing<\/a>*). While many chalk down to unrealistic users irritated\u00a0they haven’t seen their\u00a0investment quadrupling in value overnight, reading these posts\u00a0did<\/em>\u00a0make me question: what historical data in regards to returns is available for the different Acorns portfolios?<\/p>\n

Digging around, I quickly realized the answer was: not much<\/strong>.\u00a0Acorns itself publishes no info on the rate of returns of its various portfolios, which isn’t unexpected or surprising. Reading the accounts of other users doesn’t cut it for me,\u00a0since many of them bail after getting antsy or taking their first 2% loss. There’s another camp of users who claim they’ve seen some absurd rate of return (Bro, I’ve seen 65% since May!<\/em>). What about portfolio statistics? Dreaming of reading\u00a0about portfolio volatility? Keep dreaming. Point being, tangible data is hard to come by.<\/p>\n

Coincidentally, I’ve been playing around with Quantopian<\/a>, an absolutely awesome python-based algorithmic investing framework and simulation\/backtesting platform. Quantopian has access to stock metrics, such as price, volume, dividend\/split info, etc, since January 3rd, 2002; Quantopian\u00a0allows users to backtest stocks and trading algorithms\u00a0using this data, in order to examine performance of various trading strategies. Somewhere along the line, I had the thought: “Why not just create each of\u00a0the Acorns portfolios and backtest them to analyze their performance?”<\/em>. The data provided by these backtests\u00a0is generally so close to true performance for simple cases that any error is excusable.<\/p>\n

Acorns allows users to choose a portfolio based on their risk tolerance and investing goals, all\u00a0ranging from “Conservative” to “Aggressive”. The five portfolios are really just different target allocations\u00a0in six ETFs: VOO, VB, VWO, VNQ, LQD, and SHY. The first four\u00a0funds are Vanguard ETFs and the last two are iShares bond ETFs. The portfolio distributions are shown below, with the Aggressive portfolio being shown on top and the Conservative portfolio on the bottom.<\/p>\n

\"acorns\"<\/a><\/p>\n

This is basically all the info required to build these portfolios within Quantopian. Each security in the portfolios is added to the trading list, and \u00a0a rebalance function is scheduled to run every day 1 hour after the market opens. In reality, the Acorns portfolio rely on new round-ups and contributions to keep allocations at the right percentages, but this method of simulation seems close enough. I also set the backtest to only hold long positions, and made sure the trading commissions were zero (so as to only focus on the portfolio performance, and not the Acorns fee breakdown). The entire algorithm is shown below.<\/p>\n

\"\"\"<\/span>\r\nAttempts to model the Acorns Agressive Portfolio for performance modeling.<\/span>\r\nBacktesting is limited to September 10, 2010 (first trade day of $VOO).<\/span>\r\nAcorns keeps allocations exact by purchasing fractional shares. Since this isn't generally possible, a larger initial capital must be used (>$10K).<\/span>\r\n\"\"\"<\/span>\r\n\r\ndef<\/span> initialize<\/span>(context<\/span>):\r\n \"\"\"<\/span>\r\n Called once at the start of the algorithm.<\/span>\r\n \"\"\"<\/span> \r\n set_long_only()\r\n \r\n # Each security in the ETF along with the target percentage<\/span>\r\n context.voo =<\/span> (sid(40107<\/span>), 0.14<\/span>) \r\n context.vb  =<\/span> (sid(25899<\/span>), 0.25<\/span>) \r\n context.vwo =<\/span> (sid(27102<\/span>), 0.20<\/span>) \r\n context.vnq =<\/span> (sid(26669<\/span>), 0.30<\/span>) \r\n context.shy =<\/span> (sid(23911<\/span>), 0.05<\/span>) \r\n context.lqd =<\/span> (sid(23881<\/span>), 0.06<\/span>) \r\n \r\n context.security_list =<\/span> [context.voo, context.vb, context.vwo, context.vnq, context.shy, context.lqd]\r\n \r\n # Rebalance every day, 1 hour after market open.<\/span>\r\n # In reality, the Acorns app banks on additional buys <\/span>\r\n # To keep the allocations correct.<\/span>\r\n schedule_function(rebalance, date_rules.every_day(), time_rules.market_open(hours<\/span>=<\/span>1<\/span>))\r\n \r\n # Fees are zero, if you're on the \"student plan\".<\/span> \r\n set_commission(commission.PerShare(cost<\/span>=<\/span>0<\/span>, min_trade_cost<\/span>=<\/span>0<\/span>))\r\n set_commission(commission.PerTrade(cost<\/span>=<\/span>0<\/span>))\r\n \r\n \r\ndef<\/span> rebalance<\/span>(context<\/span>,data<\/span>):\r\n \"\"\"<\/span>\r\n Execute orders according to our schedule_function() timing. <\/span>\r\n \"\"\"<\/span>\r\n for<\/span> security in<\/span> context.security_list:\r\n   #rebalance to target percentage<\/span>\r\n   if<\/span> data.can_trade(security[0<\/span>]):\r\n     log.info(\"Rebalancing %s to %s percent\"<\/span> %<\/span> (str<\/span>(security[0<\/span>]), str<\/span>(security[1<\/span>])))\r\n     order_target_percent(security[0<\/span>], security[1<\/span>])\r\n\r\n<\/pre>\n
The only thing that changes between each of the Acorns portfolios is the target percentages. Everything else remains constant, which makes backtesting and examining\u00a0the different portfolios incredibly easy. Three backtest\u00a0start dates\u00a0were chosen: September 10th, 2010, September 10th 2014, and October 8th 2015. Each backtest\u00a0ended on the same date: October 7th 2016. The returns (%) for each backtesting window are summarized below.<\/p>\n
\"screen-shot-2016-10-09-at-1-04-39-am\"<\/a><\/p>\n
Upon first look, it appears that the returns are reasonable and on par for each portfolio description, and I can confirm that the returns are very close to the actual returns I’ve experienced during my usage of the app over the last year. Returns diminish fairly monotonically from aggressive to moderate as one would expect.<\/p>\n
There’s a catch though: I didn’t include the S&P500 benchmark performance in the results.<\/p>\n
\"screen-shot-2016-10-09-at-1-12-23-am\"<\/a><\/p>\n
In all the backtest windows, the S&P500\u00a0destroys<\/strong> the Acorns portfolios. Suddenly, the 10.57% short term return doesn’t seem so great after realizing the S&P500 did almost 13 percent in the same year. Take a look at the Quantopian backtest results for the\u00a0aggressive portfolio, the best-performing one of the bunch, over the last year.<\/p>\n
\"screen-shot-2016-10-09-at-1-16-59-am\"<\/a><\/p>\n
The portfolio almost exclusively underperforms the S&P500 in terms of returns. It’s also worth nothing that the drawdown during the January-Feburary time period of 2016 is worse than that of the S&P500 as well.<\/p>\n
At this point, the logical thought is “Of course the drawdown sucks, because this is the\u00a0Aggressive<\/em> portfolio.” And you’d be partially right. Take a look at the results for the conservative portfolio during the same time period:<\/p>\n
\"screen-shot-2016-10-09-at-1-21-52-am\"<\/a><\/p>\n
The drawdown during the same period is\u00a0hardly better\u00a0<\/em>than the S&P500, which fell over 3% from the starting point in the beginning of October 2015. Rather disappointing, given that the portfolio is billed as a “safer bet” option for investors. The backtest results over the last year of\u00a0all five portfolios are given below, for anyone interested.<\/p>\n
\"screen-shot-2016-10-09-at-1-16-59-am\"<\/a> \"screen-shot-2016-10-09-at-1-40-41-am\"<\/a><\/p>\n
\"screen-shot-2016-10-09-at-1-41-00-am\"<\/a><\/p>\n
\"screen-shot-2016-10-09-at-1-41-20-am\"<\/a><\/p>\n
\"screen-shot-2016-10-09-at-1-40-16-am\"<\/a><\/p>\n
My final thoughts? It’s not that these portfolios are\u00a0bad<\/strong>. They’re decent portfolios, especially for beginning investors who are looking to stash away a few pennies here and there with minimal effort and a simple fee structure. Is a 6-10% return better than holding onto a pile of cash and losing out to inflation? Absolutely<\/strong>!\u00a0An Acorns portfolio serves as an excellent “baby’s first investment”. I just can’t help but shake the disappointment that all five\u00a0portfolios have underperformed the market in most, if not all, previous financial circumstances. For anyone beyond the novice saver\/pocket-change investor, it makes more sense to\u00a0invest wisely in a few basic\u00a0ETFs (i.e.\u00a0NOBL\/UPRO<\/strong>)\u00a0and call it a day. A true zero-fee investment experience can be had by using Robinhood<\/a>, potentially saving the $1 per month fee. Smart savers who appreciate the auto round-up\u00a0feature of Acorns can enjoy Wealthfront<\/a>‘s auto-deposit feature with the added benefit of Tax-Loss harvesting<\/a>.<\/p>\n
Do you have thoughts on my analysis of these portfolios or my backtesting strategy? Let me know in the comments. I appreciate reader input.<\/p>\n
*r\/investing deserves a post of its own. Similar to r\/fitness<\/a>, the\u00a0“beginners pretending to be experts” culture leads to an unsurprising amount of\u00a0awful<\/em> advice. Trust the internet at your own risk. Trust the internet with investment guidance at your bank account’s risk.<\/p>\n","protected":false},"excerpt":{"rendered":"
I’ve been using Acorns, the app to “help anyone invest” for a little over a year, mainly as a curiosity. One common theme I’ve seen amongst Acorns users is the disappointment with the lack of returns throughout their usage. Simply searching “Acorns Returns” online brings up hundreds of posts (many of them,\u00a0uncoincidentally, occurring on r\/investing*). … <\/p>\n
Continue reading “Analyzing the Performance of Acorns Investment Portfolios using Quantopian”<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":179,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":true,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","enabled":false},"version":2}},"categories":[60,61,62,1],"tags":[63,68,67,64,69,66,65],"class_list":["post-175","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-acorns","category-investing","category-quantopian","category-uncategorized","tag-acorns","tag-backtesting","tag-investing","tag-portfolios","tag-quantitative-analysis","tag-quantopian","tag-returns"],"jetpack_publicize_connections":[],"jetpack_featured_media_url":"https:\/\/i0.wp.com\/www.blog.jay-greco.com\/wp\/wp-content\/uploads\/2016\/10\/Screen-Shot-2016-10-09-at-1.07.03-AM.png?fit=1168%2C189","jetpack_sharing_enabled":true,"jetpack_shortlink":"https:\/\/wp.me\/p755Zm-2P","jetpack-related-posts":[],"_links":{"self":[{"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=\/wp\/v2\/posts\/175"}],"collection":[{"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=175"}],"version-history":[{"count":4,"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=\/wp\/v2\/posts\/175\/revisions"}],"predecessor-version":[{"id":190,"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=\/wp\/v2\/posts\/175\/revisions\/190"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=\/wp\/v2\/media\/179"}],"wp:attachment":[{"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=175"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=175"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.blog.jay-greco.com\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=175"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}