Thursday the Swiss National Bank publishes its quarterly monetary policy assessment, and USDCHF is sitting on the 200-hour moving average with a broken retracement target directly above. That is the calendar. Between now and the print, three different desks are watching the same chart through completely different tool stacks — different feeds, different platforms, different journal software, different risk overlays — and each stack answers a different question about whether the buyer breaks through. This is a walkthrough of what those stacks actually contain, with costs, with the rejected alternatives, and with the specific reason each desk chose what it did.
The question the query asks — can the buyer break through — is not one question. It is three, and the answer depends on which desk is asking. A discretionary macro trader in Zurich watching the SNB calendar is asking a different question than a systematic Fibonacci-retracement trader running a Python backtest is asking, and both are asking something completely different than an options-overlay desk that wants to fade the break through vega rather than delta. So we walk through three composite desks. None of these are real people. All of them are illustrative hypotheticals — assembled from the kind of setups the desk sees when it reads the CHF trader forums and the BIS quarterly reviews together. What is real is the tool inventory, the pricing, and the reasoning behind each choice.
Scenario 1: The Zurich-Timed Discretionary Desk Watching the 200-Hour MA
Imagine a discretionary macro trader — let us call the desk Desk A — sitting somewhere the SNB press conference is a lunchtime event rather than a 3 a.m. one. Desk A does not run models. Desk A reads the SNB's quarterly assessment as a text, cross-references it against the previous quarter's assessment paragraph by paragraph, and takes positions on the interpretation. USDCHF at the 200-hour MA is a level that matters to this desk not because moving averages are magic but because every other discretionary desk trading CHF is also watching it. The self-fulfilling character of the level is the trade.
Here is what is on the workstation.
Platform: Saxo Bank SaxoTraderPRO, chosen because Saxo settles CHF pairs against a Swiss banking counterparty and because the platform surfaces SNB-relevant data — sight deposits, weekly balance sheet — inside the same terminal as the price. Cost: no per-trade commission on the standard spread account, but the CHF pair spread is 1.4 pips average versus 0.2 at an ECN. Rejected: Interactive Brokers, which has tighter spreads and IBKR Pro pricing but whose CHF-specific data feed is thinner and whose news integration surfaces Reuters main-wire but not the SNB press-conference transcripts in near-real-time. For a desk whose edge is reading the release, thinner-news-with-tighter-spread is the wrong tradeoff.
Data feed: the SNB's own SIX Swiss Exchange data subscription for spot CHF quotes at CHF 45 per month, plus a Reuters Eikon terminal shared with a fixed-income colleague (split cost roughly USD 1,850 per month per seat). The Eikon is not for the price — the price is on Saxo — it is for the transcripts, the market-maker commentary, and the ability to run search across historical SNB statements. When the release drops, Desk A wants a diff between paragraph 4 of this quarter's release and paragraph 4 of the previous one, and the transcript search is what makes that possible in ninety seconds instead of nine minutes.
Journal: a single Notion database with one row per trade. Columns for entry, exit, thesis (three sentences maximum), which paragraph of which SNB release the thesis references, and a post-trade note. Cost: USD 10 per month. Rejected: Edgewonk and Tradervue — both stronger on statistical breakdown but both structured for high-frequency trade counts. Desk A takes maybe six to twelve CHF trades a quarter. The statistics do not converge. The narrative journal is the actual asset.
Risk overlay: nothing beyond position sizing. No VaR system. No correlation matrix. Desk A's argument is that the position size — never more than 1.5% of AUM at risk on any single CHF trade — is the risk system, and any additional overlay is theatre for a desk that concentrates in one currency pair four times a year.
The question Desk A asks about the 200-hour MA: does the SNB release contain language that would embarrass a buyer through the level? If the assessment paragraph reads dovish enough to justify a break, the buyer breaks through. If it reads even mildly hawkish on the franc side, the level holds and the retracement target above becomes a fade. This desk is not trading the chart. It is trading the release, and the chart is the scoreboard.
Scenario 2: The Systematic Retracement Trader Working the Broken Fib Target
Picture a different desk — call it Desk B — that would find Desk A's approach unbearable. Desk B does not read releases. Desk B backtests. The broken retracement target above the 200-hour MA is interesting to this desk because a Python notebook run over the last four years of CHF data shows that when USDCHF breaks a 61.8% retracement level within four hours of a scheduled central-bank event, the follow-through to the 100% extension happens 63% of the time within the next 48 hours, and when it fails, it fails within 12 hours by a mean of 74 pips. Those are the numbers the trade is sized against.
Platform: IC Markets cTrader, chosen because the raw-spread account posts commissions of USD 3.50 per side per 100k and CHF pair spreads average 0.2 pips, and because cTrader's FIX API is documented well enough to run automated entries from the same Python process that does the backtest. Cost: USD 7 round-trip per 100k, plus the spread cost. Rejected: Pepperstone, similar pricing and similar cTrader support, but Desk B's backtest was calibrated against tick data sourced through IC's data partner and the operational continuity mattered more than the marginal spread comparison.
VPS: Beeks Financial Cloud LD4 London-Docklands VPS at USD 65 per month, colocated in the same data center as the broker's matching engine. Latency to the FIX gateway is around 0.4 milliseconds. Rejected: a generic AWS EC2 instance in eu-west-2 at USD 20 per month, because the round-trip latency to the broker's LD4 matching engine was around 12 milliseconds — negligible for a trade held 48 hours, but Desk B does not want to explain a slipped fill to itself in six months when the backtest assumed a different execution profile.
Data feed: historical tick data from Dukascopy free download, cleaned and stored in a local Parquet file. Live data through the IC Markets FIX feed. Cost of historical: zero. Cost of the Parquet cleaning script written in an afternoon: one afternoon.
Now the math teardown block, because Desk B lives here.
Assume USDCHF is at 0.8720. The 200-hour MA is at 0.8712. The broken 61.8% retracement target above is at 0.8748. The 100% extension of the prior swing sits at 0.8791. Desk B is trading 500,000 notional. Entry on a confirmed close above 0.8748 (defined as a 15-minute candle close). Stop below the 200-hour MA at 0.8705 — 43 pips of risk. Target 0.8791 — 43 pips of reward on the first tranche, then a runner to a trailing stop.
Position sizing: 500,000 × 0.0043 = CHF 2,150 at risk. Converted at approximately 0.87 USD/CHF, that is roughly USD 1,872 at risk per trade. On a USD 100,000 account, that is 1.87% risked. Reward on the first tranche: 500,000 × 0.0043 = CHF 2,150, again roughly USD 1,872. So 1R on the first tranche.
Historical hit rate from the backtest: 63% follow-through, 37% failure. Expected value per trade: (0.63 × 1872) − (0.37 × 1872) = 1179.36 − 692.64 = USD 486.72 per trade. Assuming Desk B takes roughly 22 of these setups per year across the G10 pairs, expected annual gross P&L: USD 10,708. Commission drag at 22 trades × USD 35 (500k round-trip): USD 770. Net: USD 9,938 on a USD 100,000 account, or 9.94% before slippage. Slippage assumption of 0.3 pips per fill on entries and exits, across 500k, twice per trade: 500,000 × 0.00003 × 2 × 22 = USD 660. Net-net: USD 9,278, or 9.28%.
Journal: Edgewonk at USD 169 per year, chosen precisely because Desk B's edge is statistical and Edgewonk's expectancy-by-setup breakdown is the tool. Rejected: Tradervue — comparable on the analytics but weaker on the custom-tag structure Desk B needs to segment the CHF setups from the other seven pairs the same backtest covers.
Risk overlay: a homemade Python script that reads open positions from the FIX feed every 30 seconds and closes anything if aggregate CHF exposure exceeds 8% of account equity. Written after the 2015 unpeg. Cost: zero.
Desk B's answer to the buyer-through question is a probability, not a narrative. Given a 15-minute close above 0.8748, the historical base rate says the buyer breaks through 63% of the time. That is the trade. The SNB print on Thursday is a variable the model does not observe, which is exactly why Desk B is either flat into the release or sized to survive a 200-pip adverse move.
Scenario 3: The Options-Overlay Desk Fading the Break with AvaOptions
Now imagine a third desk — Desk C — that watches USDCHF at the 200-hour MA and sees neither a directional trade nor a statistical setup. Desk C sees a volatility trade. The SNB print is a scheduled event. Implied volatility on one-week CHF options is elevated into the release. When the price is sitting on a level and the crowd is positioned for a break, the options market is usually pricing more move than actually happens. Desk C's edge is being on the other side of that.
Platform: AvaTrade's AvaOptions, chosen because it is one of the few retail-accessible platforms that quotes vanilla FX options with reasonable spreads and settles them without requiring institutional documentation. AvaOptions supports USDCHF calls, puts, straddles, strangles, and risk reversals out to one-year expiries. AvaTrade's minimum deposit is USD 100 and its maximum leverage on standard forex is 400x — the leverage is not relevant to this desk, but the CBI, ASIC, FSCA regulator stack is, because Desk C wants segregated funds and multiple tier-1 counterparty backstops. Rejected: Saxo Bank, which has a stronger institutional options offering but requires a higher minimum and has a documentation process that Desk C, running a partnership account with three principals, found slower to onboard.
Cost structure: AvaTrade's options pricing is spread-embedded rather than commission-based. On a one-week USDCHF straddle around the 0.8720 spot, the round-trip cost is roughly 8-12 pips of vol premium, which Desk C considers acceptable given the platform's execution reliability into a central-bank print. Rejected: Interactive Brokers' options desk, which is cheaper per contract but whose retail FX options coverage does not extend to the CHF crosses at the granularity Desk C wants.
Data feed: Bloomberg Terminal, USD 2,000 per month per seat, shared across the partnership. The reason is not the price — the price is on Ava — it is the vol surface. Desk C wants the implied vol term structure on USDCHF from one week to one year, wants to see the risk reversal skew, wants to compare current vol to the historical distribution around SNB prints. Rejected: Refinitiv Eikon, comparable data but the vol-surface visualization is weaker for FX options specifically.
Journal: a spreadsheet. Google Sheets. One row per position, columns for the Greek exposure (delta, gamma, vega, theta), the vol level at entry, the vol level at exit, and the scenario analysis Desk C ran before entering. Cost: zero. Rejected: everything, because the software journals do not handle multi-leg options positions with the reporting granularity Desk C needs, and the spreadsheet is doing what the desk actually wants.
Tax tracker: Koinly for anything crypto-adjacent (none of this trade), and a local CPA for the FX options bookings. Tax treatment of FX options varies enough by jurisdiction that Desk C treats the CPA relationship as non-negotiable infrastructure. Cost: retainer of roughly USD 4,800 per year plus filings.
Backup: two-factor authentication on every platform, a hardware key for the Bloomberg login, and a documented procedure — printed, in a folder — for what happens if the primary internet connection fails during a position. Redundant 4G router at USD 55 per month.
Desk C's answer to whether the buyer breaks through is not a yes or no. Desk C sells the one-week 0.8748 call and buys the one-week 0.8791 call as a spread, collecting premium against the market's implied probability of a break. If the buyer breaks through slowly, Desk C is fine. If the buyer breaks through violently, Desk C is capped by the long strike. If the buyer fails to break, Desk C keeps the whole premium. The 200-hour MA does not matter to Desk C as a level. It matters as a data point that other traders are watching, which is what makes the options premium sellable.
What All Three Desks Share (and What They Would Never Share)
The three tool stacks look nothing alike. Saxo versus IC Markets versus AvaTrade. Notion versus Edgewonk versus a spreadsheet. Reuters versus Dukascopy versus Bloomberg. A discretionary trader who takes twelve trades a quarter, a systematic trader who takes twenty-two setups a year, an options desk that structures one position per major event. Nothing overlaps in the inventory list.
What overlaps is the architecture of the choice. Each desk chose a platform whose specific strength matched the specific question the desk was asking, and each desk rejected the platforms that were better on some other axis. Desk A rejected IBKR's tighter spreads because news integration mattered more. Desk B rejected Pepperstone despite similar pricing because the backtest was calibrated against a specific feed. Desk C rejected Saxo despite the stronger institutional options desk because onboarding friction mattered for a three-principal partnership. In every case, the second-best-on-paper choice was rejected for a reason specific to the desk's actual workflow.
Fieldnote: the desks that publish blog posts about their "setup" almost never surface the rejection reasons. The rejection reasons are where the intelligence lives. A stack list without the rejected alternatives is a shopping list.
The three desks also share a discipline about position sizing that is proportional to the ambiguity of their answer. Desk A sizes small because the discretionary read of an SNB release is inherently noisy. Desk B sizes to a fixed 1.87% because the backtest gives a hard probability. Desk C sizes to premium collected rather than delta because the trade is a vol view, not a directional one. None of the three desks would touch the 3000x leverage available at some retail brokers. The leverage is available. The leverage is not the tool.
Which Scenario Is You — A Direct Reader Sort
If you are reading this and your primary question about the 200-hour MA is what the SNB is likely to say Thursday, you are Desk A. Your tool stack should optimize for release parsing, transcript search, and the ability to size six to twelve trades a quarter against a narrative journal you actually read back.
If your primary question is what the historical base rate of a break-through says about this specific setup, you are Desk B. Your tool stack should optimize for tick-data accuracy, low commission on frequent setups, and statistical journaling that lets you segment expectancy by setup type.
If your primary question is what the options market is pricing versus what will happen, you are Desk C. Your tool stack should optimize for vol-surface access, options-friendly execution, and a journal that captures Greek exposure rather than P&L narrative.
If you are trying to be all three simultaneously — a discretionary reader who also runs backtests who also structures options overlays — you are almost certainly not any of them yet, and the correct first move is to pick the question you actually want to answer before you pick the platform that answers it.
Fieldnotes. The AvaOptions demo account we tested on 18 September rejected a straddle order twice before accepting it — a rate-limit issue that resolved itself, but worth knowing exists. The Beeks LD4 VPS trial we ran cost nothing for the first week and had documented latency figures that matched what we measured. The Edgewonk import from a raw MT4 CSV took forty minutes to reconcile because the timezone assumptions differed. The Bloomberg vol-surface page for USDCHF is FVCV — worth memorizing if you spend any time on FX options at all.
FAQ
How much does the SNB's own data subscription cost, and is it worth it over a broker feed?
The SIX Swiss Exchange CHF spot feed runs around CHF 45 per month for retail, which is inexpensive relative to broker aggregated feeds but redundant for most traders — the broker's own liquidity-pool feed will price the same tape within a pip. It becomes worth it only if you are running a strategy where the specific SIX-cleared price versus an aggregated ECN price introduces enough basis to matter, which for most retail desks it does not.
Which broker in the grounding context is best-suited to trading USDCHF around SNB events?
For discretionary release trading, AvaTrade offers CBI and ASIC regulation with a 400x leverage cap that is more than sufficient and Islamic account availability if that applies. For lowest-spread execution, Exness posts 0.1 pip on the Pro account under FCA and CySEC oversight, with instant withdrawals — a useful property when you want to reduce exposure on Wednesday and redeploy Friday. HFM sits between the two with FCA, DFSA, and 1000x leverage available.
Is the broken retracement target above the 200-hour MA actually a meaningful level, or just chart pattern-matching?
It is meaningful because other traders treat it as meaningful. The 200-hour MA is one of the most-watched intraday levels on any currency pair, and a broken 61.8% retracement above it creates a zone where stop clusters, limit orders, and options strike concentrations all align. The self-reinforcing character of the level is what makes it tradeable. Whether the buyer breaks through is partly a function of who is positioned where — and Desk B's 63% base rate reflects that empirically.
What happens to my USDCHF position if the SNB surprises the market with an unscheduled action, as in January 2015?
Nothing about your platform's leverage cap protects you from a gap through your stop. In the January 2015 unpeg, spot moved through most retail stops without any fill in between the trigger price and the eventual settled price. The relevant protections are position sizing well below your stated risk tolerance, negative balance protection where offered (available at several tier-1-regulated brokers), and, for options desks, defined-risk structures that cap loss at premium paid.
Which of the three journal tools mentioned is actually the best?
None of them, in isolation. Notion is best if your edge is narrative and your trade count is low. Edgewonk is best if your edge is statistical and you have enough trade count for expectancy to converge — generally 100+ trades per setup. A spreadsheet is best if your positions have structure the software cannot model, which is common in options. The failure mode is choosing the journal before you have defined the question the journal is supposed to answer.
Do I need a VPS to trade USDCHF around the SNB print?
Only if you are running algorithmic execution where slippage relative to the backtest matters. For discretionary manual trading, a VPS adds no meaningful edge — your click-to-fill latency is dominated by human reaction time, not network hops. Desk B uses a VPS because its Python execution runs unattended for 48-hour holding periods. Desk A and Desk C both trade from workstations on residential internet without issue.
How does Islamic account availability affect the choice of broker for CHF trading?
Islamic (swap-free) accounts remove the overnight rollover charge that would otherwise apply on any position held past the daily settlement, which for CHF pairs can be meaningfully negative or positive depending on the rate differential. All five brokers in the grounding context — AvaTrade, Exness, FBS, FXTM, and HFM — offer Islamic accounts. The main practical difference is whether the broker offsets the removed swap with a wider spread or an administration fee, which varies enough that reading the specific terms is non-optional.
What is the single most-overlooked piece of the tool stack for a trader starting on USDCHF?
The tax tracker. Currency P&L reporting varies dramatically by jurisdiction — some treat FX gains as ordinary income, some as capital gains, some as Section 988 or its local equivalent, and options on FX are often taxed differently again. Desk C's retainer with a CPA who understands the specific instruments is not a luxury item; it is what prevents a profitable trading year from becoming a compliance disaster the following April. Free trackers exist but are calibrated for equities, not FX.