Tomas Vykruta Data Daily

WOW! Trump's new tax plan (2018) is taking money from single people and giving it to married couples. When did IRS become a federal marriage counselor? Take a hypothetical NYC Wallstreet earner who declares $350,000 a year taxable income. 2017 single earner keeps $206k 2017 married keeps $215k 2018 single earner still keeps $205k 2018 married earner keeps $235k! by getting married, in 2018, $30,000 tax savings. That's enough to buy a brand new luxury car _every year_, or $2500 monthly payments on a mortgage $1M home! $30,000 in _after tax savings_ is a huge amount of money in america. What the fuck? (above) tax savings for single earner. almost no savings at all! (above) Tax savings  for married earner. almost 5% lower effective federal! (above) Taxes in 2017 and 2018 for married earner. (above) tax savings for 2017 and 2018 single earner (virtually no savings).

I've spoken with numerous financially savvy people in past few months who think the long term capital gains in New York are ""somewhere 15% to 20%?". It's actually 31%-33%. If you had significant gains in, say, a company's stock, and you have a high active income, you're going to get CRUSHED when you sell. For a 400k annual active income earner your effective long term capital gains tax is 33.64%. For rich trust fund people who don't work and earn W2 income, they pay 22.45%. But of course they wouldn't be dumb enough to declare Manhattan as their home base, they'd be in Miami, Florida where the long term capital gains tax is 12.75%. Life isn't fair. don't be mad. But don't be dumb and sell your stock thinking it's a tax efficient  diversification strategy. **Take a year off work, move to Miami, and dump it there. **I'm not sure if the feds let you do this, if you earned it in NYC, they might still t...

I'm pulling all my money out of Wealthfront today. I was always pleased with Wealthfront's simplicity and performance, and today took a close look for the first time at actual performance. I was shocked to see how poorly it's actually performed. Besides poor performance, there are other reasons I'm pulling out too. Wealthfront was marketed heavily to facebook/googlers and I took it for granted that it was a great modern product, better than what the banks offer. Assumptions are the termites of relationships. 1. performance is abysmal. If I kept it in S&P, did nothing at all, In 5 years S&P is up from 1600 to 2800 or a whopping 75%. In the same time my Wealthfront portfolio did an abysmal 30%. I have a 10.0 risk score selected. 2. interface is terrible. the website heavily dedicates UI to encourage you to transfer all of your money to Wealthfront by drawing retirement graphs and showing how the more money you transfer the better off you'll be (...

Airplane maintenance is very expensive, there is no way around this. This does not mean your mechanic gets a carte-blanche. The most important advice is to not argue once a bill has been issued, the discussion must happen before the invoice, ideally, before the work is done. Mike Busch, my favorite airplane guru and mentor, recently published a Q&A on this topic. Worth watching. I have one additional tip for SR22 or IO550 owners specifically. Continental publishes a warranty-labor allowances table. You can call them directly, or look it up on their website. These are not the recommended  hours, these are the maximum allowed. It's perfectly normal to reference this guide before the work is done, and ask the mechanic if his labor hours will not exceed these amounts. If your mechanic is offended by this notion, get a new mechanic. Remember to allow extra time for decowling and recowling. It's your plane and your money, don't let other people make decisions on yo...

MYSTERE FALCON 50 Airplane Flying Manual AFM is attached here . You can download and print.

Our latest machine learning paper has been published. link to view MODELING NON-LINGUISTIC CONTEXTUAL SIGNALS IN LSTM LANGUAGE MODELS VIA DOMAIN ADAPTATION Abstract:  When it comes to speech recognition for voice search, it would be advantageous to take into account application information associated with speech queries. However, in practice, the vast majority of queries typically lack such annotations, posing a challenge to train domain-specific language models (LMs). To obtain robust domain LMs, typically a LM which has been pre-trained on general data will be adapted to specific domains. We propose four adaptation schemes to improve the domain performance of long shortterm memory (LSTM) language models, by incorporating application based contextual signals of voice search queries. Most adaptation strategies are shown to be effective, giving up to 21% relative reduction in perplexity relative to a fine-tuned baseline on a heldout domain specific development set. Initial...

Highly recommend the original white paper on crypto written by Bitcoin creator Satoshi Nakamoto: https://bitcoin.org/bitcoin.pdf Short read, not too much math, extremely well written and describes in depth not only the motivation and design, but potential vulnerabilities and why they are not practical threats. One interesting problem not addressed is explosive growth in block size. Recall that: - blocks are limited to one every ~10 mins by self adjusting difficulty - blocks must contain all transactions from previous 10 minutes Currently we are doing about 400k transactions per day, and one block = ~1MB. If we assume transaction count is roughly proportional to the amount of people using the currency, we could see 100x or 1000x growth in transaction volume. So we're looking at 100mb or 1GB blocks. This would create a bandwidth issue, every node has to download 1GB every 10 minutes, there would be propagation delays. I don't think this is a huge problem - networks are gettin...